Rapid Evaluation of Sediment Budgets

Home , Catena (soil)

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/247332796

Book · January 1996

CITATIONS READS 211 250

2 authors:

Leslie Margaret Reid Thomas Dunne US Forest Service University of California, Sa…

44 PUBLICATIONS 1,668 178 PUBLICATIONS 13,713 CITATIONS CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Modeling of hillslope evolution by water erosion View project

All content following this page was uploaded by Thomas Dunne on 17 September 2015.

The user has requested enhancement of the downloaded file. Leslie M. Reid & Thomas Dunne

RAPID EVALUATION OF SEDIMENT BUDGETS

GeoEcology paperback Die Deutsche Bibliothek - CIP Einheitsaufnahme Reid, Leslie M.: Rapid evaluation of sediment budgets / Leslie M. Reid & Thomas Dunne. - Reiskirchen : Catena Verl., 1996 (GeoEcology paperback) ISBN 3-923881-39-5 NE: Dunne, Thomas:

All rights are reserved. No part of this publication may be reproduced. stored in a retrieval system or transmitted in any form or by any means, electronic. mechanical, photocopying. recording or otherwise, without prior permission of the publisher.

This publication has been registered with the Copyright Clearance Center, Inc.

Submission of an article for publication implies the transfer of the copyright from the author(s) to the publisher.

ISBN 3-923381-39-5 Table of Contents

List of Figures List of Tables List of Boxes List of Symbols Abstract Acknowledgements

1 INTRODUCTION 1 1.1 Definition and description 3 1.2 Relation to other methods 5 1.3 Misconceptions 6 1.4 How to use this volume 7

2 PROCEDURE FOR SEDIMENT BUDGET CONSTRUCTION 9 Step 1. Define the problem 9 Step 2. Acquire background information 10 Step 3. Subdivide area 13 Step 4. Interpret aerial photographs 14 Step 5. Conduct fieldwork 16 Step 6. Analyze data 19 Step 7. Check results 20

3 EVALUATING SEDIMENT PRODUCTION FROM HILLSLOPES AND CHANNELS 20 3.1 Overview of shillslope sediment production 22 3.2 Identification of sediment sources 25 3.3 Rates of discrete erosion processes 27 3.3.1 Landslides 27 3.3.2 Debris flows 30 3.3.3 Gullies 31 3.3.4 Treethrow 34 3.3.5 Animal burrows 35 3.4 Rates of chronic erosion processes 36 3.4.1 Sheetwash erosion 36 3.4.2 Wind erosion 41 3.4.3 Dry ravel 43 3.4.4 Bank erosion 43 3.5 Other hillslope sediment transport processes 44 3.6 Sediment delivery from hillslopes to channels 47 3.7 Grain-size composition 50 3.8 Calculation of long-term rates 52 3.9 Predicting future erosion rates 59 4 EVALUATING SEDIMENT TRANSPORT AND STORAGE IN CHANNELS 60 4.1 Overview of channel processes 62 4.2 Characterization of channels 63 4.2.1 Qualitative characterization 63 4.2.2 Selection of measurement sites 65 4.2.3 Slope measurements 67 4.2.4 Channel geometry measurements 68 4.2.5 Channel roughness 69 4.2.6 Definition of flow characteristics 70 4.2.7 Identification of channel changes 71 4.3 Grain-size distributions 73 4.4 Initiation of bed-material transport 83 4.5 Determination of scour depths 90 4.6 Sediment transport rates in channels 92 4.6.1 Sediment transport equations 93 4.6.2 Comparison of transport predictions and measurements 94 4.6.3 Applying sediment transport equations 106 4.6.4 Use of field observations 113 4.6.5 Evaluation of the washload component 114 4.6.6 Limits of sediment transport predictions 116 4.7 Sediment storage 116 4.7.1 Identifying storage elements in channels 117 4.7.2 Defining trends in channel-related sediment storage 118 4.8 Computations of sediment yield 123

5 EXAMPLES OF SEDIMENT BUDGET APPLICATIONS 124 5.1 West slope of the Sierra Nevada, Californai 127 5.2 Shinyanga Region, Tanzania 130 5.3 Snoqualmie River basin, Washington 132 5.4 Olympic Peninsula rivers, Washington 133

6 CONCLUSIONS 135

APPENDIX 1. GLOSSARY 138 APPENDIX 2. ADDITIONAL READING AND USEFUL COMPENDIA 144 Background information 144 Manuals and descriptions of particular methods 146 Data compendia 147

REFERENCES 149 List of Figures

1. Simple flowchart of sediment transport on hillslopes and in channels 4 2. Definition of subareas based on bedrock and vegetation type 13 3. Sequential aerial photographs 15 4. Examples of increasingly complex flowcharts 17 5. Flowchart that indicates typical relation between sediment mobilization, production, deposition, and yield 20 6. Examples of external growth nodes 28 7. Graph of cumulative landslide volume as a function of age of the landslide 29 8. Flowchart indicating the relation of sediment transported by debris flows to sediment produced from hillslopes 31 9. Relation between gully cross-sectional area and gully width 32 10. Use of root exposure and erosion mounds to estimate surface erosion rates 37 II. Correlation between erosion-intensity class and measured recent rates of erosion in Kenya 41 12. Grain-size distribution of average annual sediment yields from 6 small catchments 52 13. Calculation of long-term landsliding rate 54 14. The error function (erf) of Urn 55 15. Landslide frequency versus road age 58 16. Components of the sediment load in a channel 62 17. Example of the relation between low-flow and high-flow water-surface profiles 69 18. Grain-size distribution of colluvium in hollows and alluvium from channels in a 4th-order basin 74 19. Grain-size distributions of channel sediment at River Mile 22 of the Snoqualmie river, Washington 75 20. Contour map of bed elevations in a meander 81 21. Percent of a channel-bed pavement that is intermittently mobile at various shear stresses 87 22. Comparisons of shear stress for initial motion of particles of various sizes predicted using initial-motion equations and those measured by Milhouse 88 23. Phases in the erosion of landslide deposits or other sediment stored behind dams of organic debris 120 24. Flow chart illustrating relationships between various processes of colluvium transport into small streams studied in the N. Fork Kings River basin 128 25. Variation along the Humptulis River of average annual rates of bedload transport 135 List of Tables

1. Recent examples of rapidly constructed sediment budgets 2 2. Common problems addressed by hillslope-based sediment budgets 21 3. Common sources of sediment affected by various land uses 26 4. Ordinal-scale classification scheme for sheetwash erosion severity 40 5. Erosion measurements on roads and paths 42 6. Examples of questions concerning channels that can be addressed by sediment budgeting 61 7. Equations for initiation of motions 85 8. Sediment transport equations A. Bedload 94 B. Total bed-material load or suspended bed-material load 95 9. Tests of sediment transport equations in natural channels A. Gravel-bedded channels 96 B. Sand-bedded channels 97 10. Histograms of modal values for sediment load equations 99 II. Potentially useful sediment transport equations for channels of various types 100 12. Characteristics of streams used to test sediment transport formulas A. Gravel-bedded channels 102 B. Sand-bedded channels 103 13. Studies that compare transport equations with data from natural channels 104 14. Expected proportions of bedload in total load 114 15. Sediment budgeting studies 125 16. Results of selected sediment budget studies 126 17. Comparisons of measured sediment yields with those calculated using sediment budgets 137 List of Boxes

1. Typical resource assessment problem 1 2. Sediment budget definition 3 3. Is a new method applicable to your problem? 8 4. Steps in sediment budget construction 9 5. Acquiring aerial photographs and maps 12 6. Evidence for activity of erosion and sediment transport processes 23 7. Indicators of surface erosion 24 8. Sample calculation of soil creep input 46 9. Recognizing sediment deposits 48 10. Calculating total sediment from a grain size trapped 49 11. Why you can't describe the world from your desk 51 12. Calculation of long-term landslide frequencies 56 13. Selecting a useful channel reach for analysis 66 14. Recognizing bankfull stage 68 15. Considerations in the use of historical evidence 72 16. Evidence of channel bed stability 75 17. A rapid field method for sieving gravel 76 18. Determining the required sample size for a pebble count 79

19. Example of a calculation of required sample size for 0 50 80 20. How comparisons are noted in Table 9 98 21. Defining the zone of active transport 107 22. Checklist for applying sediment transport equations 110 23. Evidence for channel aggradation 121 List of Symbols a constant a' factor in Hey's velocity equation A channel cross-sectional area Ab cross-sectional area of bankfull channel Aw area of catchment Awf area of catchment: future Awp area of catchment: past Agf agriculture factor in bank erosion equation b constant C constant dmax maximum depth of scour or fill in channel df cross-sectional average of fill depth in channel ds cross-sectional average of scour depth in channel D particle diameter Di particle diameter of interest Dg geometric mean of particle diameters D, diameter of tree

D50sub median particle diameter in subsurface

D IO median particle diameter

D84 particle diameter than which 84% of the grains are smaller DD drainage density erf error function E(y) expected value ofy, given probability density of fiY) f Darcy-Weisbach friction factor fly) probability density of y g acceleration due to gravity (approximately 980 cm/s2) h flow depth he flow depth at which particle begins to move h.n height of root mound h.nax maximum flow depth H, hydraulic sensitivity in bank erosion equation J percent clay in soil K soil erosivity L length sampled for estimating density of linear features Mi median of sample set i in calculation of required sample size Mr weight of roots n roughness factor in Manning equation, or number of samples or sample sets N number of samples required for given precision of diameter estimate p index of particle suspendibility q water discharge per unit stream width Q water discharge Qb bankfull discharge of stream R hydraulic radius S water surface gradient Sm average gradient of mainstem channel SO sediment delivery ratio T amount of rain of intensity> 0.5"/24 hours Tr amount of rain of intensity> 0.5"/24 hours: future Tp amount of rain of intensity > 0.5"/24 hours: past u mean flow velocity u. shear velocity of flow V root mound volume Wm width of root mound w, settling velocity of sediment grains x variable y variable

Y b bank erosion rate

Y g gully retreat rate Ygr gully retreat rate: future Y gp gully retreat rate: past e base of natural logarithms: 2.71828

~ mean ~x mean of normally distributed climatic parameter in landslide calculations It 3.14159 p density of fluid p, density of sediment cr standard deviation cr x standard deviation of normally distributed climatic parameter in landslide calculations , shear stress of flow 'c critical shear stress 'c * critical dimensionless shear stress erg reference dimensionless shear stress in Parker (1990) equation ~ scale for describing grain size, ~ = 1.44 In (particle size in mm) Abstract Many land-management decisions would be aided by an understanding of the current sediment production and transport regime in a watershed and of the likely effects of planned land use on that regime. Sediment budgeting can provide this information quickly and at low cost if reconnaissance techniques are used to evalu ate the budget. Efficient budget construction incorporates seven steps: careful definition of the problem to be addressed; collection of background information; subdivision of the project area into uniform sub-areas; interpretation of aerial photographs; fieldwork; analysis; and checking of results. Methods used in field work and analysis must be selected according to the types of hillslope and channel processes active, the goals of the analysis: and the level of precision required. Methods for evaluating erosion and sediment transport rates are described, and four examples are given to demonstrate budget applications and construction.

Acknowledgements

Many colleagues have contributed to this document by their extremely useful comments on previous drafts. In particular, we thank Lee Benda, Robert Beschta, Michael Kuehn, Mary Ann Madej, Walter Megahan, Mark Smith, Kate Sullivan, Andre Lehre, and Fredrick Swanson. We also thank James Baldwin, Stephen Burges, William Dietrich, Jack Lewis, Thomas Lisle, Robert Thomas, Jeff Cederholm, and Weihua Zhang for suggestions on particular aspects of sediment budgeting technique and applications, and Luna Leopold for his unfailing encour agement. Dunne was supported by a lS. Guggenheim Fellowship during part of the writing.

Leslie M. Reid Thomas Dunne

June 1996

Retrieval terms: watershed, cumulative impact, erosion, sediment transport, sediment budget, land-use planning, watershed analysis Chapter 1: Introduction 1

1 INTRODUCTION

Researchers and land managers are increasingly interested in the response of erosion and sedimentation to changes occurring on watershed hillslopes or in stream channels. Managers are concerned with evaluating the combined effects of a variety of land uses in a drainage basin. They need to predict how land use will alter erosion and sedimentation rates, and they need to know the relative importance of different sediment sources to assign priorities for erosion control. They also must anticipate where sediment will be deposited, how long it will be stored, and how it will be re-mobilized. Scientists interested in land-surface processes want to under stand the effects of climatic change on landscape development, determine the rela tive geomorphological effectiveness of storms of different magnitudes, understand

Box 1 Typical resource assessment problem addressed through sediment budget analysis Imagine that large-scale logging has begun to spread into a hilly region. Resulting changes to watershed conditions and processes may include reduced rainfall inter ception and evapotranspiration, altered rates of snowmelt, increased soil-moisture recharge and runoff, reduced root-reinforced soil strength, increased soil compac tion, altered hillslope gradients due to road construction, and altered drainage patterns. Resource managers may reasonably ask hydrologists and geomorphoiogists to predict the extent to which these land-surface alterations will change the hydrology, sediment loads and channel morphology of streams in the region. The geomor phologist will be responsible for the issues related to sediment, although it will be necessary to make such an assessment in cooperation with the hydrological analy sis. The resource managers are likely to ask specific geomorphological questions such as: How much fine-grained sediment is likely to wash off road surfaces and be deposited in the interstices of channel bed gravels? To what extent will landslide sediment input to streams be increased? From which parts of the landscape will such landslide sediment originate? Will any landslides evolve into debris flows and scour channels? If so, can the debris-flow paths and their zones of scour and depo sition be predicted? Will there be zones of accelerated sediment accumulation? Will there be a resultant change in channel depth, width, or pattern? How long will it take for channels scoured to bedrock to recruit new bed material and to establish a new alluvial morphology? If sediment-control strategies such as road closures or regrading were to be employed, would they significantly affect the sediment input to channels, the texture of channel bed material, or the channel morphology? These are the kinds of questions to which a well-constructed sediment-budget can contrib ute useful answers. 2 Chapter 1: Introduction the influence of hills lope transport processes on channel morphology, and unravel the environmental history recorded in sediment deposits. Sediment budgets are a useful tool for addressing these management and research problems. Many questions confronted by land managers are mandated not only by consci entious watershed management, but also by legislation. In the United States, for example, the National Environmental Policy Act requires that the cumulative effects of land use be evaluated. Cumulative watershed effects may occur either at the site ofland-use activity or downstream, and downstream effects can be caused only by a change in the input or transport of a watershed product such as water, sediment, organic material, or solutes (Reid 1993). Sediment is usually the component mos~ profoundly changed by land use, and alterations to either the water or organic com ponents also often affect sediment production and transport. In addition, altered channel morphology is a cumulative impact of wide concern, and this occurs only through mobilization of sediment. Analysis of potential or existing cumulative effects thus usually requires an evaluation of sediment inputs and transport, and sediment budgeting is a useful tool in such analyses. In other cases, particular management projects need to be evaluated with respect to on-going watershed processes. The demand for this type of analysis has risen

Table I: Recent examples afrapidly constructed sediment budgets

Location and Person-days Reference Purpose Field Office Puget Lowland, Evaluate the effects of logging and road 38 20 27 WA (Madej construction ori channel morphology 1982) Olympic Mts., W A Estimate sediment input from forests to compare 375 32 20 (Reid 1981) with input from gravel roads Shinyanga, Analyze the relative importance of sediment 21,000 41 75 Tanzania (Reid sources to aid in allocating soil conservation 1990) investments on semi-arid range and cultivated lands Machakos, Kenya Analyze the relative importance of sediment 14,000 2 15 (Reid 1983) sources to aid in allocating soil conservation investments on range and cultivated lands Snoqualmie R., Define low-head hydro impacts in a partially 212 22 24 WA (Dunne logged basin 1988) Sierra Nevada, CA Define low-head hydro impacts in a partially 80 14 40 (see text) logged basin Kauai, HI (Reid Determine effect of Hurricane Iniki on future 1325 6 6 and Smith 1992) flood hazard on Kauai Chapter 1: 1ntroduction 3 with the increasing awareness that both the cause of and the solution to most instream problems lie upstream, in the watershed. As a result, aquatic resource specialists find an increasing need for geomorphological watershed analysis, and sediment budgeting can provide useful information in this context, too. Land managers and researchers often assume that sediment budgeting is a time consuming exercise suitable only for long-term studies. However, we have routinely constructed sediment budgets for a variety of applications using an approach that usually requires no longer than two months of fieldwork and analysis (table 1). In this volume we outline the method and describe its application to specific problems. The volume is intended as a guide fbr.land managers who are contemplating the use of sediment budgets and for those who will oversee their construction. It is not intended as a how-to manual for budget construction, although we hope that the geomorphologists and hydrologists directly involved in sediment budgeting might find useful techniques described. Although methods of budget construction are relatively uncomplicated, they require that users have extensive backgrounds in geomorphology and hydrology if the methods are to be used appropriately.

1.1 Definition and description A sediment budget is an accounting Box 2 of the sources and disposition of A sediment budget is an accounting sediment as it travels from its point of the sources and disposition of of origin to its eventual exit from a sediment as it travels from its point of drainage basin (figure 1). In its full origin to its eventual exit from a form, a sediment budget accounts for drainage basin. rates and processes of erosion and sediment transport on hills and in channels, for temporary storage of sediment in bars, alluvial fans, and other sites, and for weathering and breakdown of sediments while in transport or storage (Dietrich et al. 1982). Although complete sediment budgets are of scientific interest, they are more detailed than is necessary to address the problems most frequently encountered in resource management. For most application, some combination of the following information is required: • the type and location of the major natural and management-related sources of sediment • the approximate amount of sediment contributed by each type of source • the 'grain-size distribution of sediment contributed from each source • the approximate volume and grain sizes of sediment in storage along streams • the approximate transport rate of sediment through stream channels and valley floors 4 Chapter i: introduction

:I: ;= r JJ) 5 '"C m

--- Stream bank

(') :I: > Channel Transport Z Z m r --- Basin Mouth

Figure 1: Simple flowchart of sediment transport on hills lopes and in channels. In this case, treethrow and soil creep intermittently transport sediment downslope to the channel bank, where it is eventually delivered to the channel by channel-side treethrow, bank collapse, and bank erosion. Sediment is then alternately stored and transported by the channel to the watershed mouth. Processes are noted as ovals, storage elements as rectangles, and transfers as arrows; the stream bank appears as a dashed line.

The infonnation that is relevant depends on the problem posed, and answers to the first three points often suffice. Sediment budgets may be constructed for any of several steps in the production and transport of sediment in a watershed. If analysis is focused on the potential lifespan of a soil, then the relevant budget components are the at-a-point erosion and deposition rates of soil. Lowering rates due to each erosion process and deposi tion rates from each transport process would be measured and summed to estimate the total lowering rate. In many cases, the cause for concern is the rate at which sediment enters a channel system. To address such problems it is necessary to evaluate the rate at which sediment is delivered to channels from hillslopes. For convenience, we will refer to this step as hillslope sediment production. Assessment of sediment produc tion requires infonnation about the rate at which sediment crosses the hillslope channel interface, but the rate of lowering on hillslopes can be ignored. Other studies are prompted by concern over changing channel fonn, and these require analysis of deposition and erosion rates within stream channels. Changes in sediment storage must also be evaluated if the rate ot sediment export from a watershed is to be detennined. The watershed sediment yield is usually calculated as a volume or weight of sediment leaving the watershed per unit time if rates of Chapter I: Introduction 5 downstream sedimentation are of concern, but yields are usually calculated per unit area of watershed if characteristics of different watersheds are to be compared. Empirical results from sediment budgeting studies may be transferred to other watersheds with similar climate, geology, soils, and land use. This is the use in tended for most measurements of sediment yield or hydrologic response from "representative" basins, and the transfer of sediment budgeting results is subject to the same limitations as the transfer of other types of information: processes must be well-enough understood to ensure that the basins are comparable; the significance of unusual events must be recognized; and results must be accepted only as esti mates. Sediment budgeting has often been used to describe existing patterns of sediment transport and production in a watershed, but it can also be used to predict changes in sediment regime. If rates of sediment production from various sources can be correlated with controlling variables such as vegetation cover, hillslope gradient, or rain intensity, then the effects of altered controlling variables can be estimated. For example, Corbett and Rice (1966) used measured landslide rates on grasslands and in chaparral to predict the effects of brushland conversion on erosion rates, and Reid and Dunne (1984) defined relations between traffic intensity, rainstorm size, and road-surface erosion to allow estimation of the erosional consequences of any distribution of road use in a watershed. Similar approaches might be used to predict the consequences of changes in climate, expansion of land-use activities, or alterna tive strategies for erosion control.

1.2 Relation to other methods Although sediment budgeting produces much information traditionally acquired by monitoring, construction of a sediment budget is not an alternative to monitor ing. Sediment budgeting may complement monitoring programs by providing information needed to plan measurement strategies or to interpret results, and monitoring can be used to refine sediment budget estimates if a more precise quantification is necessary. Budget construction requires identification of erosion processes and their con trols, and estimation of their rates. This information can be used to target processes for which monitoring will be most useful, to estimate the measurement precision or the number of samples necessary to characterize process rates, and to combine data and results from a variety of monitoring techniques. For example, Dietrich and Dunne (1978) used a sediment budget to identify the most important sediment production and transport processes in a coastal Oregon watershed and suggested that future monitoring be focused on those processes; and Reid (1981) used a budget to collate sediment contributions measured using many different monitoring and observational methods at a variety of scales. 6 Chapter 1: 1ntroduction

In addition, sediment budgeting allows the sediment yield to be estimated before the results of long-duration stream-sampling programs are known. The method also provides information that is more useful than point measurements of sediment yield. For example, a sediment budget can be used to estimate the range of variation in annual sediment yields from a catchment by evaluating the likely variation in inputs from each sediment source; to predict differences in yields from basins with differ ent arrays of controlling variables; or to partition a watershed's sediment yield into contributions from particular tributary basins or sediment sources. Comparison of sediment-budget results with measured sediment yields suggests that the uncertainties involved in the two methods can be comparable (see later, table 17). The crude measurement techniques used in approximate sediment budg eting often introduce errors no larger than those produced by poor temporal and spatial coverage in short-term sediment sampling or process monitoring.

1.3 Misconceptions There is a persistent misconception among both land managers and researchers that construction of sediment budgets is a time-consuming, academic exercise that is impractical for addressing the goals of land-use planning or short-term research. However, methods have been developed that greatly extend the applicability of the technique. Although sediment budgeting often uses long-term measurements, budgets can also be constructed using rapid measurements and estimates to provide results at a level of precision adequate for most management needs. Sediment budgets are mistakenly viewed as impractical for short-term analyses in part be cause the utility of approximate budgets is often overlooked. A second misunderstanding arises because erosion and transport rates are diffi cult to measure precisely, accurately, and consistently. Erosion is perceived as being intractably variable and complex, and lengthy measurement periods are assumed to be necessary if monitoring is to produce a meaningful average erosion rate. How ever, it is possible to design simple sampling schemes that account for seasonal and local variations in process rates if the reasons for these variations are understood. If landslide rates are observed to depend on rock type, for example, data can be tabulated separately for different rock types. In addition, most management appli cations require only that the order of magnitude or the relative importance of process rates be known. For example, it may be necessary only to show that land slides annually contribute 20 to 40 t/km2 in a particular basin, rather than 200 t/km2, and that they are a more important sediment source than sheetwash erosion. To many people, sediment budgeting implies construction of detailed maps of erosion processes, and this results in a third misconception. Management problems usually involve areas that are too large to permit thorough examination either in the field or on aerial photos, so comprehensive mapping is impractical and sediment Chapter I: Introduction 7 budgeting is thus assumed to be impossible. However, construction of budgets for large areas merely requires a modification of technique. Large areas are divided ("stratified") into subunits of similar soils, bedrock, vegetation, topography, and land use, and each subunit is characterized by budgets constructed for representative areas within it.

1.4 How to use this volume This volume first describes the stepswe have found to be useful for constructing sediment budgets (Chapter 2). Many of these are routine and can be carried out by those with minimal training in geomorphology and hydrology. However, planning and implementation of fieldwork and interpretation of results requires a much more sophisticated knowledge of these fields. We thus discuss methods used to evaluate rates of sediment production from hillslopes (Chapter 3) and transport mechanisms in channels (Chapter 4) in much more detail. Finally, we describe four sediment budgets (Chapter 5) to illustrate how the methods are used. Appendices provide a glossary and an annotated list of useful source books and recommended background reading. The report has two goals. First, it is intended as a resource for land managers who are confronted with particular sediment-related problems and need advice on methods available for evaluating those problems. Tables 2 and 6 identify procedures applicable to particular problems and can be used to determine which sections of the report will be most useful. If the problem is addressable by sediment budgeting techniques, then Chapter 2 should be examined to gain a general understanding of sediment-budgeting procedures, and Chapter 5 to provide examples of applications. Topics of particular relevance can then be selected from the balance of the report, which is written to provide broad overviews of the procedures available. Used in this way, the report is effectively a shopping list of available procedures, from which the subset relevant to a particular problem may be chosen. The other intended audience is geomorphologists interested in constructing sediment budgets for research or management applications. These users may find the report most useful as a guide to literature on specific analytical techniques, and as a source of ideas for applying those techniques. It would require a volume of textbook proportions to fully describe each of the procedures surveyed in this report. In most cases, we can provide only the briefest introduction to a particular technique and must direct readers to the original sources for further information. The report attempts to survey the range of techniques available, but few applications would require the use of more than a handful of the techniques mentioned. In other words, the reader should not be overwhelmed by the wealth of methods available, but should carefully select only those that are useful 8 Chapter I: Introduction

for the particular application. Readers must also be prepared to evaluate critically the methods described in the original published sources. Many techniques have limited ranges of applicability, and many have been validated by measurements over only part of their intended range. We hope that our discussions will enable readers to select useful methods from future publications as they become available and to be able to judge the validity of those techniques.

Box 3 Is a new method applicable to your problem? You will often find yourself in the position of having to evaluate the utility and validity of a new method. Some of the questions you may wish to ask include the following: 1. Is the paper in a peer-reviewed journal? If so, the authors have enough faith in the technique to submit it independent critics, and it has already survived one close examination by experts. 2. Is the method developed using data relevant to afield application? Some meth ods are based on laboratory measurements in highly controlled settings and may not address the complications expected in the field. 3. Are data points shown on graphs? Without data points, it is difficult to judge the accuracy, consistency, and range of applicability of the relationship. 4. Has the method been validated using independent field data? A method is useful only if its results can be trusted. Valid methods will have been field checked against data that were not used in the model's construction or calibra tion. 5. Was the method developed using site-specific data, or can it be generalized to other locations? Methods and models that were developed for conditions at a particular site usually cannot be applied to other sites without recalibration, and their validity must be tested using field measurements. 6. What are the method's assumptions? Read the description carefully to identify the assumptions, and check these against the conditions at your site. 7. What is the method's range of applicability? Do your site conditions fit within the range for which the method was developed and tested? 8. Is the use ofstatistics valid in development and validation of the method? 9. Are appropriate data usedfor validations and calibrations? Chapter 2: Procedure 9

2 PROCEDURE FOR SEDIMENT BUDGET CONSTRUCTION

There are as many approaches to sediment budgeting as there are sediment budgeters. However, we have found that construction of approximate sediment budgets proceeds most smoothly if it follows a Box 4 consistent series of steps, as described in this Steps in sediment chapter. budget construction The time requirement noted for each step are the 1. Define the problem range encountered for the analyses listed in 2. Acquire background Table 1. These studies encompassed areas of 38 information to 21,000 km 2 and involved varying degrees of 3. Subdivide the area complexity. The time required for constructing 4. Interpret aerial sediment budgets· for the largest area is not photographs much greater than that needed for smaller areas 5. Conduct fieldwork because sediment sources in large areas are 6. Analyze the data generally evaluated by sampling representative 7. Check results subunits (table 1).

Step 1. Define the problem The success of a project usually depends strongly on how specifically its objec tives are identified. Sediment budgeting is capable of providing such a wealth of information that it is often tempting to define more of the budget than is necessary to address the identified goals. However, most studies require only part of a sedi ment budget to satisfy their primary objectives, and the part that is needed depends on what the objectives are. The minimum precision needed to meet the objectives must also be defined. Most studies require only qualitative, order-of-magnitude, or relative results, and it is much easier to obtain these than to define precise values for process rates. It is far more cost-effective to define the essential part of the budget well than to do a mediocre job on the entire budget. Tightly constraining the scope of a study not only makes analysis easier, but also helps identify exactly what information is needed to solve the problem and aids in conveying the results to decision makers. Once the study objectives are defined, the type of sediment budget needed must be identified. The aspect of erosion, transport, or deposition to be measured depends on the use intended for the sediment budget. If the reason for concern is the effect of sediment on streams, then the sediment production rate must be defined. But if the rate of thinning of a soil profile is the major concern, then the ultimate fate of the sediment is irrelevant, and the local rates of erosion and deposition are the impor tant unknowns. Rates of input to and erosion from channel storage may be important if channel changes. are an issue, and sediment yields are often required for analysis of sedimentation, water quality, and other impacts downstream. 10 Chapter 2: Procedure

Selection of study objectives usually requires lengthy discussions with those who are planning to use the results. These discussions should cover not only what information is desired, but how that information is to be used, what types of deci sions will depend on it, and what level of detail is required for the intended deci sions. Often such a discussion will reveal that the types of information actually needed are very different from those initially identified as being important. Defini tion of objectives may require a few hours to a few days.

Step 2. Acquire background information In most cases, there will already be a lot of information available about erosion processes and rates in the project area or in areas of similar geology, climate, vegetation, and land use. Information may be in the form of published literature, agency records, studies contracted to engineering and geotechnical firms by private interests, local theses and dissertations, or anecdotal information from residents or others interested in an area. Background information can be used to review the initial problem definition and to determine which erosion and transport processes are likely to be important. Such a review aids in prioritizing fieldwork and specify ing fieldwork goals. In addition, missing information can be recognized at an early stage, and fieldwork can then be planned to supply the necessary data. Information on process rates in other areas is often useful for identifying factors that control rates in the project area and for indicating the likely relative importance of processes. This information is also invaluable for planning fieldwork. Saunders and Young (1983) compile rates of solifluction, soil creep, sheet erosion, chemical denudation, landsliding, cliff retreat, and total denudation, and NCASI (1985) describes the results of landslide surveys in the Pacific Northwest. In both cases, it is useful to examine the original data sources to ensure that site conditions are similar to those in question. Reservoir sedimentation data are often useful for checking sediment budget estimates of sediment yield. Dendy and Champion (1978) compile reservoir sedi mentation measurements through 1975 for sites throughout the continental United States, and much of this information can be updated by contacting managers of the surveyed reservoirs. Larson and Sidle (1980) describe sedimentation data and erosion studies from the Pacific Northwest. Individuals or organizations likely to have useful information should be identi fied early to determine what information is available and to arrange site visits and interviews. Anglers and whitewater paddlers can usually describe river channel changes very accurately, and farmers often have precise information on the progress of bank erosion and gUllying. Road maintenance personnel usually can estimate the frequency of landslides that affect roads in their area and can pinpoint the location and dates of culvert washouts. Local historical societies and long-term residents can Chapter 2: Procedure 11 provide information about past land-use activities and impacts, and these sources may have early photographs that allow comparison of earlier landscapes with present channel form or hillslope conditions. Once a preliminary assessment is made of the erosion and transport processes likely to be present, methods for evaluating their rates can be researched. These methods include reconnaissance field techniques, monitoring, aerial photogramme try, and use of predictive equations. The feasibility and data requirements of each can be evaluated for the project. Which evaluation methods are selected usually determines the types of background information needed. Existing aerial photographic cover!lge should be identified for the project area. The Earth Science Information Center of the U.S. Geological Survey maintains an index of imagery available for the United States, and copies are available at many university libraries. Delivery of photographs may take as long as a month, so this step must be carried out early. Topographic, vegetation, soils, precipitation, land use, and geologic maps also should be obtained at this time. Where maps are unavailable, it may be necessary to plan early fieldwork to provide at least recon naissance-level information. Alternatively, low-level air surveys can be used to sample the characteristics of a project area and describe them statistically (e.g. Norton-Griffiths 1983, Abel and Stocking 1987). Reconnaissance aerial surveys are usually carried out along transects, and can provide observer counts of features such as landslides, gullies, or structures within a delineated strip, as well as periodic vertical photographs that can be point-counted to estimate the areal proportion of different surface or land-use types. Van der Plas and Tobi (1965) describe a method for evaluating the accuracy of point-count data. If the desired sediment budget requires information on sediment transport through channels, then data on flow frequency are usually needed. The U.S. Geo logical Survey is the principal agency gathering flow records for the United States, and its records for each state are available from District offices, in the Water-Supply Paper series published by the agency, and on CD-ROM diskettes that can be pur chased from EarthInfo, Inc. (5541 Central Avenue, Boulder, Colorado 80301). State water agencies, irrigation authorities, the Army Corps of Engineers, and the Bureau of Reclamation may also have gauging records and flow duration curves for streams in or near the project area. County or regional flood control districts may have flow duration curves and detailed isohyetal maps for their areas, and reservoir manage ment agencies are likely to have flow information. If a geographic information system (GIS) is available, it may be useful to enter existing in~ormation onto the system at this point so that it is ready for later analy ses. However, data entry and verification are time-consuming tasks, and much sediment budgeting work does not require the precision that GIS can provide. The potential benefits of using a GIS should be weighed carefully against the extra time that the approach requires. Unless the information is already present on the GIS and 12 Chapter 2: Procedure

Box 5 Acquiring aerial photographs and maps The U.S. Geological Survey Earth Science Information Center (ESIC) maintains a catalog of much of the aerial photography available for the United States. The ESIC imagery catalog provides a listing of this information and is available on CD-ROM at many university libraries and USGS offices. The catalog lists the flight dates, the scale of the photographs, the focal length of the lens used, the type of imagery (e.g. black-and-white, color, infra-red, etc.), the proportion of cloud cover, and other information for each photo sequence on every USGS topographic map quadrangle, as keyed by the latitude and longitude of the south-east comer of the map. Most sites have multiple coverage. If you do not have direct access to the catalog, com municate directly with: Earth Science Information Center U.S . Geological Survey 507 National Center Reston, VA 22092 (703) 860-6045 In Canada, records of coverage are held by: National Air Photo Library Room 180, Surveys and Mapping Building Department of Energy, Mines and Resources 615 Booth Street Ottawa, Canada KIA OE9 (613) 995-4568 Often, some coverage is available that is not listed in official . catalogs. Uncata logued coverage can usually be found by talking to large land owners in the area of interest, or to companies that do aerial photographic contract work in the area. To determine what photographic coverage is available in other nations it is necessary to contact the relevant agency in those nations. In many countries, aerial imagery is considered classified information and is simply not available.

the GIS operators consider the study to be of high priority, we have found that traditional maps are more useful and efficient than GIS for most reconnaissance level sediment budget studies. On the other hand, an initial investment of time in establishing GIS coverage may be justifiable if data acquisition will continue for other studies in the same area. Acquisition of background information usually has required 2 days to 2 weeks for the studies we have worked with. The time required depends on the types of data available, the size of the area, and the precision needed. Chapter 2: Procedure 13

Step 3. Subdivide area A general precept of sediment budget construction might be: "Every project area is too large and too complex". Project areas are most readily evaluated if they are divided into subareas of relatively uniform character on the basis of attributes likely to control erosion and sediment transport: geology, soils, vegetation, land-use, and topography. Subdivision may be carried out by superimposing layers on a GIS or on transparent map overlays, as popularized by McHarg (1969) (figure 2). Aerial photographs are particularly useful for mapping the distribution of attributes, and satellite imagery can provide a crude visual integration of many attributes to facili tate subdivision of very large field ·areas. Bull et al. (1988) describe a method for defining uniform strata by using multivariate classification procedures to identify homogeneous classes, but such an approach is overly complex for most reconnais sance-level sediment budgets.

D SHALE / GRASSLAND SANDSTONE/GRASSLAND

• SANDSTONE / WOODLAND

• SHALE / WOODLAND

Figure 2: Definition of subareas based on bedrock and vegetation type. Superimposition of the geologic and vegetation maps defines subareas with four combinations of bedrock and vegetation.

Our projects have generally used 1 to 25 strata defined on the basis of topo graphy and geology, with land use .considered a treatment variable within a stratum. The number selected is a compromise between having few enough strata to allow adequate measurements of each and having enough to adequately characterize the range of conditions in the project area. Subdivision has usually required only a few hours if background data are already available. 14 Chapter 2: Procedure

Step 4. Interpret aerial photographs It is curious that many people engaged in resource assessment place so little value on the usefulness of the aerial photographic record that they use no photos, or use only a single flight sequence, or fail to procure the overlapping pairs of photos that allow increased acuity and measurement precision from stereoscopic interpre tation. Skillfully interpreted, the 50-year-Iong sequence of aerial photographs available for most areas represents one of the most valuable environmental records available. Each project described in Table 1 has used aerial photographs to identify erosion and transport processes, measure or categorize process rates, and select sites for fieldwork. Scales most useful for process evaluations are larger than 1:24,000, although scales as small as 1: 64,000 are usable if no others are available. Imagery may be selected from a particular season to optimize the visibility of a specific process or feature. Landslides and bank erosion are most visible in early spring before deciduous trees have leafed out, while areas of potential sheetwash erosion usually are most evident at the end of the dry season. Many published sources provide information on identifying and interpreting geomorphological processes using aerial photographs. Jones and Keech (1966) describe the use of air photos in water erosion surveys, Foggin and Rice (1979) and Slosson (1979) provide guidelines for recognizing landslides, Emery (1975) dis cusses identification of sheetwash, gullies, and landslides, and Way (1978) provides a more general discussion of the interpretation of substrate properties and geomor phological processes. The severity of an erosion process can be estimated in different subareas by comparing the areas affected, as measured from aerial photographs. However, measured areas must be rectified to correct for the non-vertical view angle if terrain is steep, and the potential masking effect of a tree canopy must be considered (e.g. Froehlich and Pyles 1987). Measurements of the area affected by particular pro cesses can also be used in conjunction with field measurements of sediment trans port rates or erosion depth to calculate total erosion. The progress of erosion or aggradation often can be measured by examining a temporal sequence of aerial photographs (figure 3), and these often provide the best available record of land sliding, sediment storage, and land-use history. Ideally, temporal sequences should be from the same season, and valid comparison of channel features requires that the flow conditions represented by the photographs be known. Representative areas can be sampled in each subunit if the project area is very large. However, care must be taken to ensure that the areas sampled represent a statistically valid basis for characterizing the entire project area. Roels (1985) describes problems of sample selection in process studies and notes that valid inferences can be drawn only from statistically random samples. The areal extent of a vegetation or land-use type is often measured using point counts, and Van der Plas Chapter 2: Procedu re 15

Figure 3: Sequential aerial photographs demonstrate channel migration and revegetation of gravel bars on the Clearwater River, Washington, between 1951 and 1977 16 Chapter 2: Procedure and Tobi (1965) provide a graph to detennine the number of points necessary for estimates of a given accuracy. Aerial photographs are very useful for selecting appropriate field sites for meas uring erosion and transport processes. Field sites should include at least one hill slope location in each subunit, as well as representative links of each stream order. Statistical subsamples of landslides and other major sediment sources are usually targeted for field measurement to relate source area to volume lost, estimate ages, and collect source material for grain-size analysis. Analysis of aerial photographs generally requires 2 to 15 days. The length of time required depends on the number of subunits identified, their size, their uni fonnity, and the types of processes active in the area. If landsliding is common, much of the analysis of process rates is done using photographs, while evaluation of sheetwash erosion rates requires a higher proportion of field time.

Step 5. Conduct fieldwork If the project area is readily accessible, a preliminary field tour is useful to identify active erosion and sediment transport processes, detennine what field measurements are possible, and guide aerial photographic interpretation. Otherwise, fieldwork is most efficiently done after preparatory office work is complete, be cause only then has the infonnation required to meet the study objectives been identified. The importance of well-constrained objectives is particularly evident during fieldwork: the complexity of most landscapes would demand intenninable fieldwork if process distributions and rates were to be completely defined. Field work can be reduced to a manageable amount only by detennining exactly what infonnation is sufficient to address the problems posed. Preliminary work often reveals what processes are active and also allows their relative importance to be ranked. If some processes are relatively insignificant, they might be considered only briefly during fieldwork to ascertain their order of magnitude. Work would then be focused on carefully defining the rates of the few processes that have the greatest effect. At this point, construction of a preliminary flowchart of process interactions increases the efficiency of fieldwork by identifying those processes and interactions that are least well understood. Flowcharts may be constructed for any level of complexity desired. For example, Figure 4a simply diagrams the processes responsible for contributing sediment to a hypothetical channel, while Figures 4b and 4c include additional infonnation on the sediment transport processes that recharge the delivery mechanisms and on the fate of sediment in the channel. As estimates of process rates are made, these can be incorporated into the flowchart to identify the most important processes and interactions (figure 4d). Chapter 2: Procedure 17

A. C.

W 0- Sen -1 -1 I Streambank

-1 W Z Z

B. D.

w 0- S j I

Stream bank - 40 f-='---'=--=;r==-~-:'i t5r-~~ ____~'- __-'~~ __~

-1 Channel Sediment W Z Z ~ ()

Figure 4: Examples of increasingly complex flowcharts for a hypothetical first-order basin. Processes are noted as ovals, storage elements as rectangles, and transfers as arrows; the stream bank and basin mouth are noted as dashed lines. A. Sediment delivery processes. B. Hillslope transport processes and storage, and their relation to sediment delivery. C. Sediment delivery and fate of delivered sediment in channels. D. Quantification ofsediment transfers (values in t-km"'yr"').

Specific fieldwork tasks and goals depend on the problems being addressed and the nature of the field area. Work usually includes checking results of air-photo analysis by examining samples of interpreted forms to verify interpretations and by comparing photo-based and field-based measurements to determine the accuracy of 18 Chapter 2: Procedure photogrammetric results. Air-photo measurements are often augmented by field measurements of the depths and ages of erosion scars, and identified problem areas are visited. Most erosion processes leave evidence of their rates, and such evidence is used to estimate rates at representative sites in each subarea. Estimates of process rates usually require field measurement of the volume or depth of sediment displaced and the age of the activity. The most useful measurements often result from unique features of the project area and can not be anticipated. Each process should be evaluated using as many independent methods as possible. Most reconnaissance level measurements are approximate, so multiple estimates are useful for evaluating how accurate the estimates are likely to be. Specific methods for evaluating erosion rates on hillslopes and in channels will be described in Chapter 3. Preliminary rate calculations should be carried out during fieldwork to ensure that all necessary data are being collected, to estimate the sample size that will be necessary to obtain statistically sound results, and to determine whether a particular measurement strategy will work. Background work will have identified the types of data needed to compute rates using predictive equations for erosion or sediment transport, and some of these data will need to be measured in the field. Required information for calculating water erosion rates usually includes descriptions of soil texture, vegetation cover, armor ing, and road characteristics, while estimates of channel sediment transport usually require information about channel bed-material texture, width, depth, and slope. If sediment storage is to be evaluated, the volume of sediment in temporary storage along representative reaches of each channel order may need to be estimated. Soil and channel sediment samples are often collected for later analysis of grain-size distribution. Infill rates in stockponds, reservoirs, or basins created by road berms often can be estimated to check sediment yield estimates. If fieldwork can be planned far enough in advance, then the most opportune season for work should be considered. Evidence for many processes disappears soon after their activity ends, but fieldwl)rk during the season of greatest geomor phological activity is often logistically difficult. We have found it useful to schedule work in most areas for just after the end of the wet season. Fieldwork usually requires 2 days to 6 weeks, and the length of time required depends on the number of subareas defined, the size of the area, the accessibility of the terrain, the precision required, the defined objectives, and the types of processes active. Although the researcher enters the field prepared with a battery of measure ment techncques for the processes identified, each area presents its own opportuni ties for additional measurements that will not be recognized until they are seen in the field. Extra field time thus should be scheduled to take advantage of these fortuitous opportunities. Chapter 2: Procedure 19

Step 6. Analyze data Data from maps, published reports, and fieldwork are collected at widely differ ent scales and in many formats. Map data provide areal characterizations (e.g. vegetation and soil types); transect data are in the form of continuous information along a line (e.g. gully incidence noted during a low-level aerial survey); and field measurements indicate conditions and rates at specific points. These data formats must be reconciled during analysis. A common method for combining data sets is to divide each subunit into grid cells and characterize each grid cell using average values for the point, transect, and areal data falling within it (e.g. Norton-Griffiths 1983, Abel and Stocking 1987). Alternatively, the subarea can be characterized as a whole, though this masks the range of variation present within the subarea. Some map data are tedious to convert to areal characterizations unless approximations are used. Chang et al. (1989) compare quick methods for calculating average gradient from topographic maps, and Mark (1974) provides a simple technique to estimate drainage density (DD) from the number of times (n) a traverse of length L crosses a channel: DD=1.57n/ L (2.1) This method can also be used to estimate the density of roads or other linear fea tures intersected by survey transects. Most geographic information systems can automatically calculate the density of mapped linear features. Preliminary rate calculations will have been carried out during fieldwork to verify measurement strategies, but most calculations will need to be redone after fieldwork is complete. In addition, long-term averages often must be estimated from available short-term data, and this may require analysis of weather records or land use history. Calculations using predictive equations must also be carried out, and if several appropriate predictive methods are available, they all should be used. Results from different methods can be compared to estimate the accuracy of calcu lated values. Measured rates can also be compared to published values for similar areas as a check on their reasonableness. If the compared values differ widely, possible reasons for the difference should be evaluated, and further work may be needed to reconcile the differences. Average rates are then distributed according to subunit or land use within a subunit. The accuracy of estimated rates can be assessed by evaluating the accuracy of component measurements, comparing estimates using different techniques, or testing the sensitivity of calculated results to likely uncertainties in the component data. Follow-up fieldwork is often useful at this point to resolve questions that arise during analysis. Analysis has generally required 1 to 6 weeks. 20 Chapter 3: Sediment production

Step 7. Check results Dendy and Champion (1978) and Larson and Sidle (1980) have compiled rates of reservoir sedimentation and drainage basin sediment yields for the continental United States and the Pacific Northwest, respectively. These types of data, or analogous data such as basin sediment yields calculated from suspended sediment sampling and flow records, can be used to check the reasonableness of basin yields estimated from sediment budgets. Valid comparisons require that the areas have similar geology, land use, and vegetation, and information about these characteris tics usually can be determined from maps. It is often possible to measure sedimen tation rates in topographic depressions within the field area, and this information can also be used to check budgeting results.

3 EVALUA TING SEDIMENT PRODUCTION FROM HILLSLOPES AND CHANNELS

Estimates of erosion and sediment production rates are necessary for most sediment budgets (figure 5, table 2). Where the rate of sediment input to streams is

30 25 ... MOBILIZA TlON 105 how fast is soil lost? c .<: 0 'iii '"co e 1: QJ ... PRODUCTION 80 QJ 0>< QJ C how much gets into streams? .<: co '" .0

10 15 25 5 ... SEDIMENT YIELD 55 why is the lake filling in? DEPOSITION why is the stream BASIN MOUTH aggrading?

Figure 5: Flowchart that indicates typical relations between sediment mobilization (erosion). production. deposition. and yield. Line widths are proportional to the amount of sediment trans ferred. and values are shown in t-km-2yr-'. An average of 105 t-km-2yr' are mobilized on hillslopes by erosion processes. but 25 t-km-2yr' are redeposited on the slopes. so only 80 t-km-2yr-' are produced to stream channels. Gfthis amount. 35 t-km-2yr' are redeposited in channel storage locations. while 10 t-km-2yr' are removed from these storage elements. Aggradation thus removes 25 t-km-2yr' from transport. and only 55 t-km- 2yr' are exportedfrom the basin Chapter 3: Sediment production 21

Table 2: Common problems addressed by hillslope-based sediment budgets Evaluation method* Question to be addressed R S H D G R 0 G How much sediment will enter channels? S How rapidly is soil being lost? R D G How will soil productivity change? R S 0 G What processes will be triggered by land use? S H G How much difference will erosion control make? R S 0 G On which sources should erosion control be R 0 G focused? How sensitive is this site to a change in land use? R S H G How has past use changed erosion processes? R S H 0 What activities are responsible for the present R S H D G conditions? From what sources does sediment come? R 0 G How can sediment input to channels be decreased? R S D G How long will it take to recover from these impacts? R S H 0 G

* types of analysis: H historical records and aerial photographs R evaluation of erosion rates 0 analysis of deposition S use of similar site as an analog G grain-size analysis

the major issue, the "bottom line" of a sediment budget need only be the rate of sediment production. However, if sedimentation rates in channels or reservoirs are the major concern, then sediment yields for chosen outflow points or drainage areas must also be calculated. The flowchart in Figure 5 illustrates the relationship between these quantities for a hypothetical watershed, with values reported as mass of sediment per unit area per unit time. Erosion processes mobilize colluvium on a hillslope, and sediment trans port processes intermittently move the sediment downslope or redeposit it on the slope. In this case, of the 105 t_km·2 of sediment originally mobilized each year, 25 t_km·2 is redeposited. The soil delivered to the base of the slope is "produced" to the stream network. The proportion of this sediment influx that contributes to the sedi ment yield at the basin mouth depends on the grain-size distribution of the sediment supplied from each source. In the example shown in Figure 5, all the fine-grained colluvium supplied by sheet erosion leaves the basin, while only 50% of that contributed by gullying and 20% of that from bank erosion reaches the basin mouth 22 Chapter 3: Sediment production without redeposition. The rest of the sediment is at least temporarily deposited in the channel or on floodplains. If the channel is in equilibrium, then the amount of sediment eroded from floodplain deposits equals the deposited volume, but if the floodplain is aggrading, there will be a loss of sediment to long-tenn floodplain storage (25 t-km· 2_yr· 1 in Figure 5). Erosion from all sources, less deposition, then equals the amount of sediment yielded from the watershed (55 t-km·2_yr· I). Yield calculations thus require information both about sediment production and about in- channel transport and storage. Whether the desired quantity is a sediment production rate, a deposition rate, or a sediment yield, selection of measurement strategies requires a basic understanding of how sediment is mobilized and transported on hillslopes and of how it is contributed to channels.

3.1 Overview of hillslope sediment production Although some sediment may be imported to a watershed by aeolian deposition, most is produced by weathering of bedrock within the watershed. During this process, a relatively high proportion of the original bedrock mass is removed by dissolution, while the remainder remains in place as less-dense, weathered saprolite (Dietrich and Dunne 1978). Physical and biological processes may dislodge both bedrock and saprolite, and gravity usually imparts a net down-slope component to this motion if the material is on a slope. Once displaced, weathered bedrock and saprolite are called colluvium. In-situ saprolite and displaced colluvium together make up the hillslope regolith, which is roughly equivalent to the material referred to as "soil" in most discussions of erosion. Colluvium may be mobilized by landslides, sheetwash, soil creep, root growth, treethrow, burrowing, and many other disturbances. Some mobilizing processes can transport sediment all the way to a stream channel, but, more often, mobilized colluvium is soon redeposited on the slope and remains in storage until another transport event occurs. Sediment production to a stream channel occurs when colluvium or bedrock is transported from hillslope to stream. The production rate is evaluated as the sum of the rates of colluvial bank erosion and sediment transport across channel banks. Because the channel bank is a well-defined interface, it provides a convenient datum for comparing rates. It is important to distinguish between stream banks composed of colluvial sediment and those fonned in alluvium. Alluvial sediment has previously been produced to and transported by the channel and is now in channel-related storage, so remobilization of this sediment cannot be counted again as sediment production if the rate of input to storage equals the remobilization rate. Only where the removal of sediment from storage is a result of an altered sediment regime can this sediment input be included as new sediment production. For example, Pleistocene outwash deposits are relics of a channel regime that no longer Chapter 3: Sediment production 23 exists. Rates of aggradation and erosion are no longer in balance wh~re channels mine these deposits, and alluvial bank erosion can be counted as sediment production. The same is true of imbalances caused by recent land-use changes. If rates of alluvial bank erosion have increased without a corresponding increase in deposition rate, then remobilization of alluvial material must be included in the sediment budget. Such an imbalance occurs if a river is undermining a high bank while depositing a new floodplain at a lower level. This chapter describes methods for evaluating sediment transport on hillslopes and sediment production to channels. These methods are usually sufficient for constructing sediment budgets that address questions of erosion and sediment

Box 6 Evidence for activity of erosion and sediment transport processes Hillslope processes Wind erosion Dust, sand ripples, deposits in the lee of obstacles Rainsplash Erodible pedestals capped by pebbles or twigs Sheetwash erosion Muddy surface flows, small· sand fans, sorted deposits upslope of organic debris on hillslopes Rilling Rills, small sediment fans Gullying Gullies, deeply incised channels Tunnel erosion Soil pipes, sediment fans, disappearing flow; collapse pits in swales Animal burrowing Soil mounds, holes Trampling Footprints, terracettes Treethrow Fallen trees, root mounds Dry ravel Loose sediment below bare faces or above organic debris on hillslopes Soil creep Tilted fence posts; ubiquitous process, but often very slow Landslides: Rockfall Discolored patch on cliff face, talus accumulation Rock slide Rock-bound scar, rocky debris mound Debris avalanche Shallow planar scar, hummocky debris mound Earthflow Irregular hillslope surface, lobate deposit Slump Hummocky hillslope, arcuate scar, rotated blocks Excavations (human) Linear or regular mounds or depressions Channel processes Debris flows In-channel debris fans, poorly sorted massive sedimentary deposits, channels scoured to bedrock Fluvial transport Sand and gravel bars, sorted stratified sediment Dredging Deposits in stream-side dumps, linear channels Gravel mining Pits in floodplain 24 Chapter 3: Sediment production production. Sediment budgets for evaluating aggradation. sedimentation, water quality, and sediment yield usually require additional information about the transport and storage of sediment in channels. and the following chapter describes methods for evaluating these components.

Box 7 Indicators of surface erosion Water erosion (no single factor is diagnostic except #1) 1. Observation of the process in action ' 2. Mounds of relic soil around plants and under pebbles 3. Gravel lag deposits 4. Discolored lines around stones where soil has been removed 5. Accumulations of transported sediment in depressions and above obstacles 6. Sediment fans at slope breaks 7. Presence of rills or gullies 8. Observation of muddy water on the hillslope surface 9. Exposure of roots (usefulness depends on plant species) 10. Color and texture change on bark near base of plant II. Unscorched collars around rocks after a fire 12. Lichen-free bands at the base of rocks 13. Exposed parent material 14. Bare spots in grassland: these are likely sites to look for other evidence 15. Uneven topsoil thicknesses (may appear as light and dark patches) 16. Accumulations of surface litter (often as small dams) 17. Areas bared of surface litter Dry ravel (no single factor is diagnostic except # 1) I. Observation of the process in action 2. Accumulation of loose rock, sand, and litter at the base of steep slopes, in slope depressions, and above obstacles 3. Bare spots on steep. loose-surfaced slopes Wind erosion (no single factor is diagnostic except # 1) I. Observation of the process in action 2. Mounds of relic soil down-wind of obstructions 3. Lag deposits of gravel and sand 4. Accumulations of transported sediment down-wind of obstructions 5. Dust storms, blowing dust 6. Parallel furrows in clay surfaces; especially if trending across slope 7. Ripple marks on sandy soil 8. Presence of dunes 9. Asymmetric plant development (suggests wind activity) 10. Appearance of "sand-blasting" on up-wind rock faces and fence posts II. Exposure of roots (usefulness depends on plant species) Chapter 3: Sediment production 25

3.2 Identification of sediment sources Processes that produce sediment can usually be identified by walking along streams of various orders and noting the occurrence of bare soil close enough to the stream to contribute sediment. Similar surveys along hillslope transects reveal erosion processes active on slopes. Gleason (1953), Anderson (1974), and UNEP (1978) describe indicators of sheetwash and wind erosion, such as mounds of relic soils around plants and under pebbles, gravel lag deposits from which finer soils have been removed, discolored lines around stones that indicate the depth of soil removed, and accumulations of transported sediment in depressions, above obstacles, and in small fans. In each case, evidence must be examined carefully to ascertain that the features are erosional: mounds may also form by deposition, lags may be a normal feature of pebbly soils, and discolored lines may merely represent the depth of leaf litter. Indicators are most reliable when more than one is present, and the best indicator of all is observation of the process in action. Slosson (1979) discusses geomorphological features associated with unstable slopes, and Pole and Satterlund (1978) describe plant indicators characteristic of deep-seated slumps and earthflows. Wieczoreck (1984) provides a comprehensive review of indicators of hillslope instability. If sediment production is the quantity of concern, hillslope toes must be examined to determine whether some of the mobilized sediment is redeposited. For example, part of a landslide's debris usually drapes the hillslope between the source and the channel. Areas of deposition often can be recognized by sediment accumulations around plants, since many plants record their original rooting base by a change in stem morphology (e.g. butt swell on some trees), the location of roots, or bark texture. Measurable deposition may occur over duff layers where accumulation is rapid. Deposited sediment also may be recognized by its texture and stratification. If soil loss is the major concern, representative hillslope profiles should be examined to define the distribution of erosion processes along the slope. Erosion is usually particularly severe at some slope locations, while others characteristically undergo aggradation (e.g. Dunne et al. 1979). These locations depend on topography, vegetation, and the type of process. After processes of erosion, transport, and sediment production are identified in the field, flowcharts again provide the most useful way to summarize linkages between sediment sources, transport processes, and storage units. The preliminary flowcharts should be revised to reflect field observations. Examples of detailed flowcharts are given by Dietrich and Dunne (1978, p. 205), Lehre (1982, p. 70), and Swanson and Fredriksen (1982, p. 131), but much simpler flowcharts suffice for many problems (see Figures 1 and 4). Flowcharts readily reveal points at which sediment transport processes change character, so they help prevent inadvertent 26 Chapter 3: Sediment production double-counting and they disclose interactions likely to affect the interpretation of measurements. Many techniques can be used to evaluate mobilization rates once the erosion processes are identified and their relationships to one another are understood. Erosion processes are classified as either discrete or chronic. Discrete processes (e.g. landslides and treethrow) occur at a specific site at a specific time and thus can be counted, while chronic processes (e.g. sheetwash erosion and dry ravel) occur repeatedly at the same sites (table 3).

Table 3: Common !J>Ources ofsediment affected by various land uses. Source Examples of methods and Sediment source type Methods for evaluation measurements Landslides natural d field survey, air photos Kelsey 1982 road-related d air photos, field survey Swanson & Dymess 1975; Reid et al. 1981 logging-related d air photos, field survey O'Loughlin 1972 Earthflows: natural d field survey, air photos Kelsey 1982 Debris flows natural d field survey, air photos Reid 1981 road-related d air photos, field survey O'Loughlin 1972; Reid et al. 1981 Treethrow: natural d field survey Reid 1981 Gullies road-related d air photos, field survey Walter 1985 grazing-related d air photos, field survey Reid 1990 agricultural d air photos, field survey Reid 1990 Bank erosion natural c root exposure, field survey Reid 1981 road-related c root exposure, field survey Reid 1981 Surface erosion: general road sidecast c air photos, field survey Reid et al. 1981 roadcuts c root exposure Megahan et al. 1983 Sheetwash natural c USLE, erosion mounds Wischmeier & Sm ith 1978 road-related c USLE, erosion mounds, Darrach et al. 1978; Farmer & runoff sampling Fletcher 1977; Reid & Dunne 1984 logging-related c USLE, erosion mounds Dissmeyer 1976 grazing-related · c USLE, erosion mounds Dunne et al. 1978 agricultural c USLE, erosion mounds, lag Wischmeier & Smith 1978 Wind erosion grazing-related c wind equation, mounds Skidmore & Woodruff 1968 Animal burrows natural c field survey Imeson 1976

*c = chronic source, d = discrete source Chapter 3: Sediment production 27

3.3 Rates of discrete erosion processes Rates of mobilization of colluvium from discrete sources are evaluated by documenting the distribution of sources in a representative area over a specified duration and measuring the average volume and texture of sediment contributed by each. Measurements of the bulk density of the mobilized colluvium allow the resulting volume-transport rates to be expressed as mass per unit time and area for comparison to other transport and sedimentation rates. The distribution of discrete sources often can be defined relative to controlling variables, such as gradient and lithology. These relations then allow the frequency of the erosion events to be predicted in other areas from the distribution of controlling variables there. The effective sampling duration is usually the most difficult parameter to define. Most erosion processes are affected by land use, so it is useful to stratify measurements by land-use type. The basic methods for evaluating process rates are independent of the type of land use the processes are associated with. However, particular land-use activities often provide additional opportunities for assessing rates, so budgeters should be alert for measurement opportunities that become evident during air-photo analysis and fieldwork.

3.3.1 Landslides Two general types of landslides are common. Debris avalanches, rockslides, and rockfalls are rapid failures that usually move only once, while most slumps and earthflows move intermittently over periods of months to centuries. Most landslides are readily identified on aerial photographs, so the photos can be used to measure slide frequency over large areas. Field measurements must be used to determine the minimum size of slides visible on photographs, to estimate the frequency of smaller slides, and to estimate the frequency of slides hidden by trees. Aerial photographs usually permit measurement of landslide areas, and these can be converted to volume estimates by applying relations between slide area and depth or volume measured in the field for each slide type (Simonett 1967). Areal measurements in steep terrain may need to be corrected for angle of view, hillslope gradient, and aspect relative to the photo center. These corrections are most easily made by measuring the apparent hillslope length on the photograph, measuring the corresponding actual length on a topographic map, and adjusting the landslide length measured on the aerial photograph by the ratio between actual and apparent hillslope lengths. Rates of shallow landsliding usually can be measured by comparing the distribution of landslide scars on sequential aerial photographs. Alternatively, the effective sampling period for a field-based survey can be estimated by dating the slides wherever possible and selecting the maximum age of the dated events. Dating is usually made possible by determining the age of vegetation growing on scars or 28 Chapter 3: Sediment production damaged by the event. Dynesius and Jonsson (1991) compare several methods of dating fallen trees, Maeglin (1979) describes methods for counting tree rings, and Campbell (l98 1) suggests modifications for making rings more visible. Trees associated with landsliding and other geomorphological disturbances often have "diffuse-porous" wood, and these have the most difficult rings to discern; we have had some success with applying food dye to cores collected with an increment borer to make rings more evident. Maeglin (1979) notes that most cored trees develop disease, so cores should be taken only when necessary. Many species set external growth nodes annually, and their approximate ages can be estimated simply by observing their growth pattern (figure 6).

Figure 6: Examples ofexternal growth nodes

Because recovery rates depend on the size of an erosional scar, the effective sampling durations may differ for different sizes of landslides. Although it is possible to define these differences (e.g. Reid 1981, pp. 146-161), this is unneces- Chapter 3: Sediment production 29 sary for most applications. The short resolution period for small slides is often immaterial because a few large slides often dominate long-term inputs. A graph of cumulative landslide volume against age will reveal if such an analysis is needed. Figure 7 shows such a plot for landslides at a site in coastal Washington. In this case, landslide scars of the sizes that dominate sediment input remain easily identifiable until they are 30 to 50 years old, as indicated by the relatively linear trend to the plot. Field observations demonstrated that many of the small slides are no longer visible after 10 to 20 years, but they contribute such a small proportion of the total landslide input that their absence in the 20- to 50-year age class affects the estimated total volume by only 15% (the cumulative 50-year volume projected from the 0- to 20-year trend is 1800 m3). If the shorter resolution period for small landslides did significantly decrease the average rate calculated by dividing total volume of slides younger than 50 years by the 50-year "sampling interval", the rising limb of the curve would be concave downward. Earthflows represent a longer period of transport and sediment production. Air photos can be used to estimate transport rates if surface features such as trees, fences, or individual slide blocks can be tracked through sequential photographs and if the depth of the flow can be estimated. Maximum sediment discharge from the flow is then estimated as the cross-sectional area multiplied by the average surface velocity.

.--. '"'.s 1800 III . 1000 "- 0 For slides less than years old: :;::; Volume =35.5 Age - 7.06 ell r 2 = 0.98 'S 200 E :::l 0 60 80 100 120 140 Age of landslide (years)

Figure 7.' Graph of cumulative landslide volume as a function of age of the landslides for natural landslides in parts ofthe Clearwater basin, Olympic Peninsula, Washington 30 Chapter 3: Sediment production

If the dependence of landsliding rates on controlling variables is understood, then landslide risk may be mapped from the distribution of the controlling variables. Factors controlling landsliding rates usually include bedrock type, topography, and land use. The same factors are commonly used to delineate subareas during budget construction, so calculation of landslide rate by stratum usually accounts implicitly for differences in controlling variables. Rice and Pillsbury (1982) use discriminant analysis to define more closely the dependence of landslide rates on controlling variables, and this approach may be useful where landsliding rates must be defined precisely. Burroughs (1984) analyzes the mechanics of landsliding to develop a process-based screening system for landslide hazard in Oregon, and Wilson (1985) describes a more subjective evaluation of controlling variables that still allows estimation of landslide hazard over large areas. Precipitation character strongly controls the temporal distribution of landslides. Sequences of aerial photographs often disclose the effects of large storms or periods of land-use activity and allow erosion intensity to be correlated with the magnitude of particular triggering events. Nilsen et al. (1976) use this type of analysis to identify rainfall characteristics capable of triggering landsliding in the San Fran- cisco area. Caine (1980) uses published records of landslide-triggering storms to define a threshold rainfall for shallow landslides, but notes that the relation cannot be used as a general predictor for landsliding because local factors control whether a slide will actually occur when the threshold is surpassed. A local analysis of the magnitude of triggering events is particularly important where an abnormally large storm has recently caused widespread landsliding. In these cases, an average of recent landsliding rates would greatly over-represent the long-term average. Meth- ods of correcting for this bias are described in a later section. Shallow landslides are usually more sensitive to characteristics of individual storms, while deep-seated landslides respond more to seasonal rainfall characteristics.

3.3.2 Debris flows Debris flows form when debris avalanches enter steep, low-order channels and scour away the sediment that has accumulated there. Landslide and channel sediments are then transported downstream as a slurry of mud and debris that usually comes to rest in a log-jam or debris fan where the channel gradient decreases. Because the flows scour channel banks and construct debris fans, recent debris flows often can be recognized on aerial photographs even in forested areas. Much of the volume of sediment in debris fans represents remobilized channel sediments that have been produced by earlier hillslope sediment transport (Reid 1981, pp. 134-138; Benda and Dunne 1987), and this volume must be subtracted if sediment production is to be estimated (figure 8). Only the volume of the original landslide and the volumes of colluvium, saprolite, and bedrock scoured from the Chapter 3: Sediment production 31

W 0... o Bank erosion --l Bank erosion Other sediment (f) by debris flows by fluvial production --l --l processes procfilsses I 30 Streambank -- 40 10 20 10 10 20 Storage 5 in channels --l 10 W Z Z« 10 I () Debris flow transport

2nd order ____ _ basin mouth

Figure 8 - Flowchart indicating the relation of sediment transported by debris flows to sediment produced from hillslopes. In this hypothetical example, sediment is introduced to channels from hills lopes at an average rate of 90 t_km-2yr-1• Of this, 20 t-km- 2yr- 1 is associated with debris-flow activity (10 t_km- 2yr- 1 from bank erosion and 10 t-km- 2yr- 1 from associated landslides). However, debris flows transport twice that volume because they also remobilize 20 t-km- 2yr- 1 from deposits in the channel. This sediment has already been delivered to the channel by other processes. Processes are noted as ovals, storage elements as rectangles, and transfers as arrows; the stream hank and basin mouth are noted as dashed lines.

channel margins contribute to sediment production from debris flows. The volume of channel deposits mobilized by the flow often can be reconstructed from observations of sediment depth in neighboring channels. Because debris-flow frequency usually decreases with increasing channel order, volumes of remobilized channel sediment are conveniently calculated per unit length of channel of particular orders. Debris flows in channels of third or higher order rarely produce new sediment, but simply redistribute sediment already stored on the valley floor. Estimates of the volume of channel sediment that can be remobilized by debris flows are useful not only for sediment-budget calculations but also for predicting sedimentation impacts downstream of potential landslide sites. For some problems, the episodic nature and individual volumes of debris flows may be more important aspects of the sediment budget than their long-term contribution to the sediment yield (Benda and Dunne 1987).

3.3.3 Gullies Most lowland gullies in unforested areas are visible on .aerial photographs with scales larger than 1:20,000, and air photos provide a convenient means of mapping 32 Chapter 3: Sediment production gully distribution. A graticule loupe can be used to measure gully widths, and these measurements can usually be translated into volumes using field measurements of the relation between width and cross-sectional area (e.g. figure 9). The U.S. Soil Conservation Service (1977) has found that widths of active gullies are typically about 3 times their depth in cohesive materials, but only 1.75 times their depth in non-cohesive sediments. This relation also varies with the age of the gully, because continued wall retreat progressively increases the width-to-depth ratio after headcut extension ceases. The activity levels of gullies are usually discernible on photographs because of differences in vegetation cover or wall slope, so separate relations can be defined for active and revegetated forms (e.g. Crouch 1987).

10 N- E • -» 8 ::J • OJ • '0 C1l 6 ~ C1l C1l c 4 • • 0 • tl • Q) C/) • • I 2 , C/) • C/) ....0 • 0 0 0 2 4 6 8 10 12 14 Gully width (m)

Figure 9: Relation between gulzv cross-sectional area and gully width for active gullies in Shinyanga Region. Tanzania. Source: Reid 1982.

Gullies triggered by road drainage on steep slopes may be difficult to see on aerial photographs, and they are often confused with small landslides. In these settings, road segments may need to be examined on foot to measure the frequency and size of drainage-related gullies. Where gullying occurs beneath a forest cover, low-order drainages must be examined in the field to determine the distribution and frequency of gullying. Sequential photographs are invaluable for measuring rates of headcut migration and for determining when gullying began. Thompson (1964) and Seginer (1966) compare headcut retreat rates measured on aerial photographs with basin variables to define predictive relations for retreat rates. In the absence of precipitation data, Chapter 3: Sediment production 33

Seginer (1966) found gully retreat rates (Yg , in m/yr) in southern Israel to be well- described by relations of the form: (3.1)

where Aw is basin area in km2, C is a constant that varies with the locale, and b is a constant that equaled 0.5 in each of the cases tested. Thompson (1964) incorporates soil and climatic variables into a similar equation: ~ = 0.0066A~49S0J4T074fOO (3.2) where Yg has units of meters per unit time, Aw is the drainage area at the gully head in km 2, S is percent slope of the valley floor above the gully head, T is the amount of rain (mm) falling at intensities greater than 13 mm/day over the unit time, and J is the percent clay in the soil. The relation is based on 210 measurements made in seven areas. More recently, the U.S. Soil Conservation Service (1977) provided a simpler equation for predicting future rates of retreat (Ygf) from measurements of past retreat rates (Ygp):

0.46 0.20 A"f (Tf J Y =Y [- - (3.3) gf gp A 1 T wp p

where A »f IA "p is the ratio of the average drainage area of a given upstream reach to the average drainage area of the reach through which the gully has moved, and Tf ITp is the ratio of the expected long-term average annual rainfall from 24-hour rainfalls of 13 mm or greater to the average annual rainfall of those intensities for the period (if less than 10 years) in which the gully head has already moved. A similar approach can be used to modify Equations 3.1 and 3.2 to allow prediction of future rates. The most appropriate equation to use depends on the types of data available and on which variables are important in a particular locale. However, no equation should be applied until enough is known about the gUllying to ascertain that the equations' assumptions are satisfied in that location. It would make little sense to use the equations to predict gully growth rates at sites where there is a change in substrate character, land use, or vegetation immediately upstream of the present gully head, or where the gullies formed during a single extreme storm. Gully outlines can be mapped on sequential large-scale aerial photographs and compared to estimate sediment input rates per unit time, again using measured relations between width and depth. Blong et al. (1982) use this method to distinguish the relative importance of headcut retreat and sidewall erosion. If volume input is calculated per unit distance of headcut retreat, equations of the form of 3.2 and 3.3 can be used to predict future sediment input rates from gUllying. If this approach is used, however, it is essential that the equation be calibrated using measured values for the project area both to ensure that the form of the relation is 34 Chapter 3: Sediment produCtion appropriate for the site and to calibrate the equation for the local conditions. Where aerial photographs are not available for the period of interest or where additional resolution is required, local residents often can provide information on retreat rates by noting when particular structures, fields, fencelines, and roads began to be affected by the gully. Gillespie (1981) uses such information to reconstruct the activity of gully erosion in Bango Creek, Australia, and similar evidence has been used to track the expansion of arroyo networks in the southwestern United States (Cooke and Reeves 1976).

3.3.4 Tree thro w Treethrow is a common mechanism for erosion, soil transport, and sediment production in forested areas, yet its geomorphological role has rarely been measured. Schaetzl et al. (1989) review literature on uprooting, and note that treethrow mounds and pits cover an average pf 20% of forest floors in the United States and Canada. The significance of treethrow varies with location on the landscape, soil type, vegetation type, and forest structure. Day (1950), for example, notes that saturated soils produce higher rates of windthrow. Riparian zones are particularly susceptible to uprooting because of steep slopes, bank erosion, and wet substrates, so uprooting can be important for producing sediment to streams even in areas where it is not a major component of soil transport on hillslopes. Rates of transport by treethrow are relatively easy to measure. Denny and Goodlett (1956) counted the number of mounds in a representative area, estimated their maximum age at about 300 yr, assumed that their centers of mass were displaced downslope one-half the mound width, and calculated their volume (V) from the equation:

_ 21t (2hlll + W,II) 2 V - 3 2 hili (3.4) where hili is mound height and Will is mound width. The product of displacement and volume, divided by the period of record and the sampling area, then provides an estimated transport rate. Similarly, Reid (1981) estimates the rate of sediment production by treethrow at a site on the Olympic Peninsula by measuring the relation between tree diameter and mound volume and subtracting out the volume of roots in the mound, as estimated using the bulk density for wood and an equation developed by Santantonio et al. (1977) for Douglas-fir:

2226 M r = I111D, (3 .5) where Mr (kg) is the weight of the roots and D, (m) is the diameter of the tree at breast height. Reid categorized blown-down trees by the extent of decay using a classification similar to that presented by Maser et al. (1988, p. 15). Vegetation Chapter 3: Sediment production 35

growing on root mounds was dated to determine characteristic ages for different decay classes. The sediment volume contributed per unit time was then calculated from the observed decay classes and diameters of downed boles. Decay rates vary considerably with microenvironment, tree species, and region, so the ages of decay classes must be established independently for each project area. In one case, Spaulding and Hansbrough (1944) report a two-fold decrease in decay rate where wood was waterlogged. Dynesius and Jonsson (1991) describe methods for dating uprooted trees. Additional soil displacement is caused by water erosion and ravel on surfaces left bare by treethrow, and this component can be evaluated using techniques described below for sheetwash erosion and dry ravel. Where sediment production is the issue, only the volume of rootwad sediment contributed to the channel system must be assessed, and average displacement is not important.

3.3.5 Animal burrows Fortunately, burrowing usually represents only a small component of soil transport, and most of the sediment transported is not contributed directly to channels. This component is usually of concern only to those interested in detailed sediment budgets and downslope transport of colluvium over long periods of time. If sediment production is the quantity of interest, then the volume of burrow mounds lying within the channel margin must be measured as a function of channel order. Most burrowing animals avoid the seasonally damp areas around stream banks, but mountain beavers, muskrats, and other riparian dwellers may be important at such sites. Burrowing activity is often seasonal, and measurements are most useful if they are made soon after the burrowing season. The ages of mounds can generally be estimated from vegetation growing on them. Transport rates due to burrowing are more difficult to estimate than production rates because little is known of the morphology of burrow networks. Some workers have assumed that tunnel distribution is symmetrical relative to each tunnel en- trance, and so have used the distance between the center of the hole and the center of mass of the mound as the average displacement (e.g. Voslamber and Veen 1985). Thorn (1978), in contrast, estimates tunnel length and volume by connecting lines between mapped gopher holes. Andersen (1987) notes that 13% to 59% of the soil excavated by observed gophers was redeposited underground in unused tunnels. Accurate assessment of transport by burrowing would require excavation of tunnel systems to determine the origin of transported sediment. A maximum displacement can be estimated by assuming all mounded sediment originated from a single tunnel extending straight upslope from the mound. At most sites, this calculation demonstrates that bUJTowing rates are small enough to disregard for most purposes. 36 Chapter 3: Sediment production

3.4 Rates of chronic erosion processes Chronic sediment sources, which contribute sediment repeatedly and are usually distributed over relatively wide areas, are evaluated by measuring or calculating process rates at representative sites. These average rates are then used to characterize the rest of the subunit. It is often tempting to monitor a chronic erosion process during a short period of fieldwork: erosion pins can easily be emplaced in roadcuts, and grab samples can be taken to measure suspended sediment loads. However, these measurements are useful only if they can be interpreted in the context of seasonal and annual variations in process rates. Erosion rates on roadcuts vary seasonally by more than an order of magnitude, for example, and a month's data can thus yield a 10-fold over- or under-estimate of the average annual rate. Dunne (1977) describes simple monitoring techniques for a variety of erosion processes.

3.4.1 Sheetwash erosion Sheetwash erosion can not exist in the absence of overland flow, yet a surprising number of erosion studies apply sheetwash erosion equations indiscriminately across the landscape. A first step in evaluating sheetwash erosion is thus to identify the types of sites subject to overland flow. Overland flow is common on roads and other compacted surfaces where infiltration rates are lower than maximum rain intensities, and these are the only sites that regularly generate overland flow in many forests. Overland flow may also be important on some rangelands, cultivated lands, mined lands and other disturbed terrain, as well as in some burned wood- lands. Overland flow that occurs because rain intensities are higher than infiltration capacities is referred to as "Horton overland flow". "Saturation overland flow" occurs where saturated soils can absorb no more water. Dunne et al. (1975) describe features characteristic of sites that generate saturation overland flow, including their swale-like topography, gleying in soils, and mesic vegetation, and noted that the susceptible area increases seasonally or through storms as saturation spreads. Even where gentle gradients and a dense vegetation cover protect such sites from rapid erosion, they are often source areas for particulate organic material and chemical pollutants. Symptoms of sheetwash erosion are described by Gleason (1953), Anderson (1974), and UNEP (1978), but the best indicator is observation of the process in action. Evidence is usually destroyed soon after flow ends, so it is wise to schedule fieldwork for seasons when the subtle surface manifestations of erosion are likely to still be visible. Surveys of erosion indicators can be combined with estimates of the areal extent and severity of each indicator. The US Department of Agriculture has developed· a standard procedure for computing rates of sheetwash erosion on croplands that requires information on rainfall characteristics, topography, soil characteristics, vegetation, and surface Chapter 3: Sediment production 37 annoring. Wischmeier and Smith (1978) describe the use of the "Universal Soil Loss Equation" (USLE) and provide a metric conversion for the equation. Cover density is an influential variable in the equation, and this usually must be measured in the field. Winkworth et al. (1962) compare methods for estimating plant cover during field surveys, Williams (1981) evaluates the accuracy of density estimates from photographs, and Evans and Love (1957) describe a rapid procedure for sampling densities by step pointing. Abel and Stocking (1987) estimate cover density using aerial photographs calibrated by field measurements. When using the USLE to estimate sheetwash erosion rates over areas such as grid cells, it is tempting to apply average values for hillslope gradient and length for each area. However, the dependence of erosion rate on these factors is strongly non- linear, so the method will not provide valid results unless the area is very uniform in character. Julien and Frenette (1986) evaluate this effect in the Chaudiere watershed in Quebec and find that the estimated erosion rate for lumped areas of up to 0.125 km2 is 1.13 times that calculated using smaller areas of uniform slope. This relation will not apply to other locales because it depends on the texture of the landscape and spatial distribution of land uses, but it does demonstrate that the magnitude of errors introduced by this effect can be appreciable even when small aggregated areas are used. USLE calculations thus should be carried out for the variety of site types present in an area and results combined according to the areal frequency of those site types. Perhaps because of its unfortunate name, the USDA sheetwash procedure has received considerable criticism demonstrating that it is not "Universal," and that local complications can invalidate its predictions. However, by understanding the physical basis for the procedure, its range of calibration, and its intended use, those applying it can recognize its limitations and avoid them or take them into account (Wischmeier 1976). The USLE method has been applied to rangelands (Blackburn 1980, Savabi and Gifford 1987), forest lands (Dissmeyer and Foster 1981), construction sites (Wischmeier and Meyer 1973) and roads (Darrach et al. 1978, Farmer and Fletcher 1977). We have found the equation to be useful for indicating important sources of sheetwash sediment and assessing their sensitivity to changes in surface characteristics even in areas where we have not been confident of the numerical predictions of soil loss. The Water Erosion Prediction Procedure (WEPP; Lane et al. 1988, Nearing et al. 1989) is now being developed to take the place of the USLE. This procedure is intended to be more process-based than the USLE, and it will be applicable to cropland, rangeland, and forested land. Independent versions are being developed for hillslopes, field-sized watersheds, and GIS-based grid cells; the first of these is currently being field-tested by the U.S. Soil Conservation Service and is scheduled to be distributed to SCS offices in 1995. Both the hillslope and small-watershed versions are expected to be. PC-based expert systems, and currently no provision is 38 Chapter 3: Sediment production planned for non-computer-based versions. The GIS-linked module will probably require a workstation, and it is likely to be interfaced with the GRASS GIS now being used by the SCS. Erosion equations are most useful for: 1) identifying the relative importance of erosion from different areas and types of land management; 2) identifying areas and management practices most susceptible to future sheetwash and rill erosion; and 3) providing approximate estimates of rates. Precise estimates of erosion rates are less defensible, but these are usually not necessary. Whenever an empirical erosion equation is used, field measurements or observation-based estimates of process rates are useful as a check on calculated values. USLE predictions often can be checked by measuring the depth to which roots have been exposed around datable vegetation and dividing that depth by the age of the plant (figure 10). Carrarra and Carroll (1979) refine this method by using changes in ring and wood morphology to identify the onset of erosion. Care must be taken to assure that the rooting base of the plant actually represents the original ground surface, but this can usually be established by observing plant morphology on uneroded sites or by consulting with botanists.

A. Root Exposure

Likely retreat since construction 7 of road cut ~

Likely retreat since germination of seedling B. Erosion Mound

Likely surface lowering over age of shrub ]..)

Figure 10: Use ofroot exposure and erosion mounds to estimate surface erosion rUles. A. Measure ment using root exposure. B. Measurement using erosion mound Specific suggestions for avoiding errors with this method are given by Dunne et al. (1978, p. 134), Chapter 3: Sediment production 39

Roots often are not exposed by erosion because the plant canopy protects the area immediately underneath. In these cases, the height of the erosional mound can be measured using methods described by Dunne et al. (1978). Care mu:;t be taken to establish the origin of the mound, since some plants form mounds by accumulation of leafy debris or sediment. True erosional mounds can usually be distinguished by their composition (mineral soil, as opposed to duff accumulations), texture (coarse sediment typical of local soils, as opposed to the sorted sediments usually contributed by deposition), and structure (lack of stratification or buried litter). Daniels et al. (1985) were able to map the extent of sheetwash erosion on oxi- dized soils of the North Carolina Piedmont by mapping the distribution of soil hues using a Munsell color chart: A horizons are progressively thinned by erosion and incorporate increasing amounts of orange clay from the B horizon when plowed. Lewis and Lepele (1982) also base estimates of erosion severity on soil depth, but they use the depth of the maximum clay content as the datum and map erosion intensity as the inverse of the remaining soil thickness above the datum. Norton (1986) performs a similar analysis using the spatial variation of loess thickness in a watershed. The date of the onset of erosion may be difficult to establish using these approaches, but if residual soils remain intact on undisturbed lands nearby, contrasts may simply be attributed to cultural effects and the date of land-use change may be used. Other reconnaissance methods make use of the accumulation of immobile, coarse sediment on the soil surface as finer sediment is washed away. Kalisz (1986) notes the increased proportion of coarse ·clasts left by removal of 2 cm of fine sediment from soils on previously cultivated Appalachian fields, and Zachar (1982; p.173) describes methods for comparing the texture of original and residual soils. Reid (1990) measures the depth of a sand lag as a function of time since fields were abandoned to estimate erosion rates through time. Measurements for these calculations included an estimate of the sand content of the soils, the average depth of the lag deposit, and the time since the field was last plowed. This approach allows estimation of the change in erosion rates through time, which permits an estimate of the recovery rate. Recovery rates can be rapid after disturbance. Megahan (1974) reports exponential decreases in erosion rate through time as several disturbed surfaces in Idaho recovered. When using lag deposits to calculate erosion rate, care must be taken to ensure that the coarse layer is actually caused by removal of fines and not simply by their redistribution into pores, and that the deposit is not a natural feature of the soil profile. Even if absolute measurements of sheetwash erosion rates are not feasible, it is usually possible to evaluate qualitatively the severity of erosion at a site. Dunne et al. (1981) develop a classification scheme for erosion severity on rangeland sites in Kenya (table 4). It was found that a group of technicians with proper training could all score a site within ±0.5 units of each other using this scale, and that the scores 40 Chapter 3: Sediment production were strongly correlated with measured erosion rates. It is possible to rapidly characterize wide areas either on the ground or from the air using this approach. However, the relevant indicators must be identified independently for each project area. Table 4 is thus presented as a template to be modified or replaced as appropriate for specific forms of sheetwash erosion in a region.

Table 4: Ordinal-scale classification scheme for sheetwash erosion severity in rangelands of Kenya (from Dunne et at. 1981). Class Description o No sign of erosion. Soil surface often rough on scale of aggregates Some signs of weak erosion, such as wash marks on the soil surface and small erosion mounds or root exposures around trees and bushes. (See Dunne et al. 1978 for a review of the care necessary for distinguishing erosion mounds from those of other origins) 2 Erosion mounds and tree-root exposures as in Class I, but some signs of more intense wash and sediment transport in minor flow concentrations, which may exhibit ripples and a few millimeters of deposited sediment upslope of large stones or dams of vegetation debris. 3 Well-developed root exposures, sometimes with surviving erosion mounds, and the ground shows general signs of intense washing, such as streaks or armor layers of sand and heavy minerals, ripples, and redeposited sediment. Root exposure even on some annual plants. 4 General signs of intense erosion as in Class 3 but with obvious flow concentrations. The surface is corrugated by grooves parallel to gradient, and the amplitude of the ridges may be up to about 10 cm, but their wave lengths are highly variable. Although the intervening grooves show obvious signs of flow concentration, their cross-sections tend to be rounded and they typically do not have channels with sharp boundaries. 5 The hillslopeis dissected by rills and small gullies with definite margins. The intervening surface is fairly smooth and gently sloping but shows general signs of intense wash and large erosion mounds or root exposures if suitable plants are present. 6 The surface is intricately dissected by gullies into a badland topography. The walls of the gullies are steep and frequently rilled. They may also exhibit: erosion pedestals under stones; dry ravelling; and soil slumps after the development of tension cracks if the gully margins are nearly vertical.

To develop a rating system for erosion severity in an area, one needs to first walk extensively through the area to observe carefully the physical condition of the soil surface. Only when these patterns are understood for the particular project area should the scale be developed. It will be noticed that biological characteristics, such as vegetation type or cover density, are excluded as classification variables: these are important controls on the rate of erosion, but they are not erosion. Because both vegetation characteristics and erosion rates respond to a variety of parameters, it is not always true that soils with the sparsest cover are eroding fastest. Physical effects of erosion thus make the best indicators, and associations between erosion intensity and vegetation cover density or type can be quantified later. Chapter 3: Sediment production 41

25,----,-----r----,-----~----,_--_, ->-'- E 20 E - ro'- 15 c o '00 10 e "0 ~ ~ 5 cc ~ • OL-__~~~~ ____~ _____L ____ ~ ____~ o 234 5 6 Erosion class

Figure 11: Correlation between erosion-intensity class and measured recent rates of erosion in Kenya (from Dunne et at. 1981).

In the example described (table 4), measured erosion rates correlate well with the ordinal scale (figure II). If approximate loss rates can be measured at enough sites to demonstrate that they are relatively consistent within each class, they can be applied to characterize each class, and thus to provide an estimate of total loss rate. Direct inter-regional transfer of calibrated values should not be attempted because of the importance of local rate-controlling variables. Road-surface erosion is a special case of sheetwash erosion that usually must be evaluated differently because of its distribution and importance. Sheet erosion on native-surfaced and gravel-surfaced roads often is a major source of fine sediment in forested landscapes. The USLE is difficult to apply on road surfaces because soil parameters and flow lengths are hard to define, and evidence for surface lowering is usually obliterated quickly by traffic and road maintenance. In addition, fine sediments are continually resupplied to the road surface by the pumping action of traffic on wet substrates and by weathering of surfacing gravels. Reid and Dunne (1984) use sediment rating curves to calculate sheet erosion rates on gravel roads bearing different traffic intensities, and it is occasionally possible to measure dateable accumulations of road-derived sediment in ponds or behind obstructions. Measure- ments from a variety of sites show a consistent pattern of surface loss rates that depend strongly on traffic levels when normalized by rainfall (table 5).

3.4.2 Wind erosion Soil loss from wind erosion is difficult to distinguish from sheetwash erosion in some areas, both because the effects are similar and because land bared by wind 42 Chapter 3: Sediment production

Table 5: Erosion measurements on roads and paths. Source: Reid (/989).

Road use (gravel % Rain Loss rate2 Location depth, cm) Slope mm/yr Soil' t/(ha-yr-cm) Reference Washington abandoned 9 3500 st-c1- 0.004 Reid & Dunne (30) 1m 1984 North Carolina light (0) 5 2000 ? 0.8 Swift 1984 North Carolina light (5) 8 2000 cl 0.5-1.0 North Carolina light (5) 10 2000 sd 0.8-1.6 North Carolina light (15) 5 2000 sd 0.06-0.12 North Carolina light (15) 6 2000 eI 0.3 Washington light (30) 9 3500 st-el- 0.03 Reid & Dunne 1m 1984

Machakos, Kenya moderate (0) 4 900 ? 004-0.93 Reid, unpublished data Machakos, Kenya moderate (0) 14 900 ? 1.0-2.83 Shinyanga, Tanzania moderate (0) ? 800 ? 004-0.93 Shinyanga, Tanzania moderate (0) 1 900 sd-Im 0.83 Shinyanga, Tanzania moderate (0) 3 900 sd 1.1-1.43 Shinyanga, Tanzania moderate (0) 3 800 sd 1.03 Washington moderate 9 3500 st-el- 004 Reid & Dunne (30) 1m 1984 North Carolina heavy (0) 5 2000 ? 2.3 Swift 1984 North Carolina heavy (5) 10 2000 sd 1.6 North Carolina heavy (5) 8 2000 cI 204 North Carolina heavy (15) 5 2000 sd 0.2 North Carolina heavy (15) 6 2000 cl 1.6 Washington heavy (30) 9 3500 st-el- 2.3 Reid & Dunne 1m 1984

, Soil texture abbreviations: cl=elay Im=loam sd=sand 2 Values for loss rate are in tons/(ha-yr) per cm of rain. 3 Bulk density of 1.2 assumed erosion is particularly susceptible to later sheetwash erosion. This distinction is not important if the rate of soil loss is the major concern, but if the potential effects of conservation measures or sediment yields to streams are to be determined, then the surface lowering rate must be partitioned by cause. Anecdotal reports of blowing dust are usually common in areas subject to wind erosion, and past activity can often be recognized by bedforms, streaking, and oriented deposition that are associated with the prevailing direction of strong winds rather than with the hillslope gradient. A procedure similar to the USLE was developed to evaluate wind erosion on croplands in the midwestern United States (Skidmore and Woodruff 1968). The Chapter 3: Sediment production 43 method has since been expanded to include evaluations of microrelief (Potter et al. 1990), computation of rangeland soil loss (Lyles and Allison 1980), and calculations for periods shorter than a year (Bondy et al. 1980). Lyles (1983) provides values of the wind climatic factor and its monthly distribution for sites throughout the western United States. A process-based wind erosion evaluation procedure is being developed for use on personal computers and is expected to· be available in the mid-1990s (Argabright 1991).

3.4.3 Dry ravel The progress of dry ravel may appear simply as a lowering of the ground surface on steep, sparsely vegetated slopes, and so may be confused with the effects of sheetwash erosion. Distinction between these processes may not be important if both occur together and loss rates are measured directly, but identification of the active processes is crucial if sheet erosion is to be estimated using a predictive equation. Ravel is capable of moving larger particles than sheetwash erosion, and debris tends to accumulate in small, high-gradient (>20°) talus cones and sediment fans, while sheetwash usually forms fans of lower gradient. Ravelled sediment may temporarily choke small channels from which it is flushed during large storms. The highest ravel rates usually occur during periods of freezing and thawing or wetting and drying, or after fires that consume organic debris barriers such as fallen logs or branches on hillslopes. Rates are usually lower when the surface is continuously damp. Near-vertical streambanks and roadcuts are particularly susceptible to ravelling. Measurements of the exposure of tree roots have been used to estimate rates of ravel on roadcuts (Megahan et al. 1983, Reid and Dunne 1984). Reid (1989) estimates the rate of ravel from the bank of an ephemeral channel by measuring the volume of loose sediment that had accumulated at the foot of the bank during the 3 months since the stream had dried.

3.4.4 Bank erosion Examination of banks along a variety of stream orders discloses the distribution and types of bank erosion processes and usually reveals the factors that control their distribution. In general, bank erosion rates are expected to increase downstream. Hooke (1980) finds that the erosion rate for alluvial banks in Devon depends on the type of bank material, and that rates increase with the square root of basin area. Hooke (1979) also shows that erosion by corrasion increases with peak discharge at the study site, while bank toppling rates depend more on the antecedent precipitation index. Similarly, Wolman (1959) observes that the most rapid rates of bank retreat occur on wet banks during floods. 44 Chapter 3: Sediment production

Bank erosion along large channels often can be recognized on sequential aerial photographs, and this method has been used in conjunction with field measurements of bank heights to estimate sediment production rates (e.g. Lehre et al. 1983). Pizzuto and Meckelnburg (1989) use air-photo measurements to explore the influence of different riparian tree species on bank retreat rates. However, the sediment mobilized by bank erosion is usually alluvial on the lowland channels where such an approach is possible, and erosion of alluvial sediment does not count as sediment production unless the eroding deposits are relics of earlier conditions. Even so, it may be necessary to estimate erosion rates of alluvial deposits if sediment storage is to be evaluated in a watershed. Dickinson et al. (1989) develop an equation for predicting bank erosion rates (Yb, cm/yr) for channels in Ontario:

1;; = 2 X 1O-lo(K25 A~2) + 1.75(05I H,) (3.6) where K is the soil erodibility factor from the USLE. Agf is a numerical index of agricultural intensity described by the author, and varies from 1 for woodlots to 45 for the most intensive agriculture; pastures may rate as high as 15 and row crops as low as 16. H, is the ratio between the critical flow depth for initiation of bed material transport and the bank-full flow depth, and serves as an index of hydraulic stability. The equation is not likely to be directly applicable to other areas, but it indicates the major factors to consider in estimating erosion rates or evaluating the applicability of measured rates. We have consistently found that streambank erosion of colluvium in low-order channels is the most difficult sediment production process to evaluate, and it usually is an important component of sediment budgets. If bank erosion occurs as discrete failures, then it can be evaluated by the approach used for landslides, but with rates calculated per unit length of each channel order to account for differences in channel form and in the erosiveness of flows. If erosion is in the form of more continuous spalling or ravelling, loss rates often may be estimated by measuring the depth of root exposure on datable riparian vegetation. Even if only a few such measurements are possible, they can be used to establish the order of magnitude of the sediment production rate. For example, Reid (1981, pp. 167-169) finds the rate of spalling indicated by each of four recovered spalls to be between 30% and 125% of the mean rate measured using erosion pins on similar banks. LaMarche (1966) uses measurements of root exposure on trees flanking a Utah channel to estimate a channel erosion rate over an 800-year period, and is also able to discern a change in erosion rate from alterations in growth rates of the trees.

3.5 Other hillslope sediment transport processes In addition to the erosion processes described ab9ve, a variety of processes operate on the soil profile to transport it gradually downslope. These processes are Chapter 3: Sediment production 45 collectively known as soil creep, and they include infilling of root tunnels and animal burrows, frost heaving, expansion and contraction by wetting and drying, and plastic flow. In each case, slight displacements are given a net downslope component by gravity. Saunders and Young (1983) compile creep rates measured at sites with a variety of bedrock, vegetation, and climate characteristics, and these measurements can be used to estimate the range of rates likely for a particular project area. Measured values are generally low, so uncertainties from transferring data between sites and environments usually have only a minor influence on the overall sediment budget. Rates may be higher in areas of clay-rich soil, and rapid creep rates should be suspected where earthflows are present or where fenceposts are tilted downslope. At many sites, tree trunks are bowed downslope near their bases. This occasionally may be evidence of rapid creep rates, but it more commonly is caused by snow loading or accumulations of fallen litter or simply by the phototropic response of young trees growing on a shady hillslope (Phipps 1974). Soil creep rates are difficult to assess indirectly. Kotarba (1986) was able to estimate creep rates by measuring the displacement of lines of ancient megalith monuments in Mongolia, and Fleming and Johnson (1975) estimate rates by measuring displacement of a fenceline. Reid (1989) estimates long-term rates of sediment delivery from hillslopes by measuring the volume of sediment accumulated at the margins of dated valley-fill deposits and subtracting out the known rates of other processes active at the site. Reneau and Dietrich (1991) measure the volume and carbon-14 age of colluvium trapped in bedrock hollows to obtain estimates of colluvial transport rates for 4,000- to 15,000-year periods in the Oregon Coast Range. If long-term rates of soil creep are estimated from soil accumulation rates, the proportion of the sediment that is trapped must be known, as well as the relative contributions of different transport processes. Estimated soil creep rates are often used to check estimates of bank erosion rates in colluvium. Sediment is assumed to be supplied to the bank by soil creep from upslope, and the rate of sediment supply to the bank is assumed to equal the rate of erosion from that bank. Supply is then calculated as the creep rate applied to the relevant area of colluvial bank. Because little is known of the vertical velocity profile of soil creep, one rarely knows the bank area appropriate for calculations. Average creep-supply rates to channel banks can be approximated by measuring colluvial soil depths on channel banks at randomly selected points, applying the estimated creep velocity to the soil depth at each point, averaging the results, and applying the average to the length of colluvial bank per unit area. The randomly selected points should include the variety of site types present along the channel margin, including recent landslide scars where soil is absent. 46 Chapter 3: Sediment production

Box 8 Sample calculation of soil creep input Compilations of creep measurements in similar soils show average transport of 10 cm3jyr per cm of hillslope width to a depth of 1 m. Field measurements: measure the depth of colluvial soil along the channel bank for the channel orders of interest (in this case, orders 1 and 2). Measurement locations are determined beforehand using a table of random numbers, and are listed as paces from the previous measurement site. Because soil creep extends only to a depth of 1 m in this case, "creep depth" is calculated as the depth of soil present up to a depth of 1 m. Measurements located on bedrock or alluvial deposits are tabulated as 0 depths. Order 1 Order 2 Paces Bank Creep Paces Bank Creep from last height (m) depth (m) from last height (m) depth (m) 25 0.5 0.5 3 1.7 1.0 13 0.8 0.8 75 1.9 1.0 16 0 0 79 0 0 15 0 0 40 0 0 34 1.3 1.0 37 0 0 93 1.2 1.0 7 0 0 59 0.2 0.2 95 2.1 1.0 36 1.0 1.0 33 0 0 34 0.7 0.7 29 0 0 80 1.8 1.0 15 0 0 57 0 0 4 0 0 44 1.3 1.0 79 1.2 1.0 22 0.8 0.8 34 1.5 1.0 65 1.2 l.0 7 l.5 1.0 39 0.4 0.4 81 0 0 17 0.7 0.7 42 0 0 25 0.1 0.1 62 0 0 04 0 0 57 l.6 1.0 52 0.8 0.8 58 1.3 1.0 40 0.7 0.7 76 0 0 average: 0.6m average: 0.4 m

cm3 60cm cm3 The average input is: 10--x --= 6.0-- for first-order channels cm·yr 100cm cm·yr

10~ x 40cm = 4 . 0~ for second-order channels cm·yr 100cm cm·yr (continued) Chapter 3: Sediment production 47

continued: Box 8 Using the measured drainage densities of 4.0 kmlkrn2 for first- and 2.3 kmlkm2 for second-order channels in the area, inputs are: Order 1: cm 3 km 1m3 100,000cm m3 6.0-- x 4.0- x 2 (banks) x 3 x = 4.8--=-2- cm . yr km 2 1,000,000cm km km . yr Order 2:

~ 2 3 km 2 b ks) 1m3 100,000cm = 18 m3 4 .0 x . x (an x 3 x . 2 . ern· yr km 2 1,OOO,OOOcm km km . yr

3.6 Sediment delivery from hiIIslopes to channels Much of the sediment mobilized on hillslopes is redeposited on the slopes, so erosion rates are not equivalent to rates of sediment production to streams (figure 5). The ratio between the amount of sediment contributed to the channel and that mobilized on the hillslope is called the delivery ratio for the hillslope. Delivery ratios are best determined from field measurements. Sediment deposits are usually visible in the field where redeposition is important, although evidence becomes less clear as deposits age. Ratios for discrete processes generally can be estimated by measuring scar volumes and volumes of sediment remaining on slopes for a representative selection of dated events. Ratios for chronic processes are often more difficult to assess unless the age of the deposit can be established. If the transport process is size-selective (e.g. sheetwash erosion), then comparison of grain sizes of the sediment source and redeposited sediment can indicate which sediment sizes are likely to be trapped. The proportional representation of the most efficiently trapped grain sizes in the source material then can be used to estimate the total sediment lost, and therefore the delivery ratio, Delivery of sediment eroded from road surfaces is closely controlled by characteristics of road drainage. Delivery is high where culverts empty directly into channels and decreases with the length of slope the runoff must cross before entering a channel. Trimble and Sartz (1957) and Haupt (1959) develop relations to predict the proportional delivery of road-derived sediment as a function of flow distance and gradient. Once their applicability to a particular site is field-checked, relations of this type can provide estimates of delivery from similar sources. An equation developed by Tollner et al. (1976) can be used to calculate the rate of deposition more precisely where sheetwash is filtered by' vegetation. The USLE was designed to calculate sediment delivery at the bottom of uniform fields and does not address redeposition along a non-uniform hillslope. If deposi- 48 Chapter 3: Sediment production

Box 9 Recognizing sediment deposits On the basis ofsediment character: 1. Mounds of stratified sediment 2. Layers of well-sorted sediment 3. Buried clasts showing fluvial rounding 4. Non-stratified mixtures of coarse, angular clasts in a matrix of finer sediments On the basis of topography: 5. Flat-bottomed depressions, "ponds" of sediment 6. Presence of lower-gradient footslope at base of steep, straight slope 7. Lower-gradient fan-shaped cones downslope of hills lope hollows 8. Hummocky terrain or lobate topography at the toes of slopes 9. Accumulations of sediment upslope of obstructions 10. Presence of other depositional landforms: deltas, gravel bars, etc. On the basis of stratigraphy: 11. Sediment above soil horizon of different texture (caution: may be an erosional lag deposit) 12. Buried leaf litter 13. Chaotic soil profiles (e.g. coarse, unweathered sediment atop a well-developed soil profile; buried soil profiles) 14. Surface layer of sediment of contrasting color 15. Presence of buried surficial material (e.g. trash, horseshoes) On the basis of vegetation: 16. Extension of subaerial-type bark on stems below soil-surface level (depends on type of plant) 17. Roots emanating from buried stem (depends on type of plant) 18. Presence of vegetation that requires droughtier conditions than those currently experienced by the site (e.g. non-riparian trees growing within the channel margin)

tional conditions vary along the hillslope of interest, application of the equation will require some evaluation of sediment delivery. Khanbilvardi and Rogowski (1984) · try adding basin-scale delivery equations (Maner 1953, Roehl 1962, and Williams and Berndt 1972) to the USLE and find that the Williams and Berndt equation provides the best predictions of measured sediment delivery ratios (SD):

SD = o40 63S//I . (3.7) where SI/I is the average slope of the mainstem channel. Khanbilvardi and Rogowski (1984) also note that ratios vary considerably with storm size. The Modified USLE Chapter 3: Sediment production 49

Box 10 Calculating total sediment loss from a grain size trapped Example: Clasts larger than 1 mm in diameter make up 20% of the soil (bulk density = 0.9 g/cm3) exposed on a 200-m2 face in a road-side ditch. A deposit of 1.6 m3 of sediment larger than 1 mm has accumulated at the mouth of a culvert in the 3 years since the road was constructed, while all but 0.8 m3 of finer grains have been removed. The road is paved, all sediment is contributed by erosion of soils exposed in the roadside ditch, and the bulk density of the deposit is 1.2 g/cm3• Since these grain sizes make up 20% of the soil,

3 1.6 m3 of coarse sediment represents 1. 6m ·1200 k~ = 9600kg of soil 0.2 m 9600kg kg or an average loss of = 3200- 3yr yr 2.4 m3 have been trapped, representing (2.4 x 1200 = ) 2900 kg of soil, so an average of 2200 kg/yr of soil is removed completely, for a lowering rate on the exposed face of: kg m3 1 m 2200-.- -.-- 2 =0.012 - yr 900kg 200m yr Procedure: 1. Examine the deposits to determine: a. what is the smallest size consistently caught? If a size is consistently trapped, it will notbe present in deposits farthest from the source. b. what is the volume of sediment of that size and larger in the deposits? c. what is the volume of sediment of smaller sizes in the deposits? d. what is the bulk density of the deposit? 2. Examine the sediment source to ensure that sediment is coming predominantly from a single source and that all grain sizes are moved as far as the measured deposit, and measure the bulk density ofthe source. 3. Determine the percentage volume of grains of the consistently trapped size and larger in the source material. 4. Determine the length of time during which erosion has been active. 5. Calculate the mass of source material represented by the volume of trapped grains of the consistently trapped size and larger by dividing the trapped volume by their original proportional representation in the source material and multiplying by the bulk density of the trapped material. (continued) 50 Chapter 3: Sediment production

continued: Box 10 6. Calculate the loss rate by dividing this value by the erosion duration. 7. Calculate input to the channel by subtracting the total mass of deposits from the total mass of source material removed, and the annual input rate by dividing this mass by the erosion duration.

(Williams 1975) uses a storm runoff parameter in place of the USLE's rainfall erosivity factor to compensate for this effect, and it produces better estimates of sheetwash erosion yield on a per-storm basis than the other USLE-based models tested. Ebisemiju (1990) regresses measured sediment delivery ratios (calculated from measurements of erosion and deposition along slope transects) against the physical characteristics of several erosion plots and reports that delivery is most closely correlated with gradient (positive ) and infiltration rate (negative) on bare plots, but is best predicted by slope length (negative) and soil erodibility (positive) on vegetated surfaces. Sediment delivery ratios from hillslopes thus depend on the competence and capacity of the transport process and on the ability of the hillslope surface to trap and store sediment. Delivery ratios are small where sediment transport capacity decreases downslope. This is particularly noticeable where hillslopes are buttressed by long, low-gradient, well-vegetated footslopes. If redeposited sediment is observed during fieldwork, its relation to factors such as gradient, surface roughness, vegetation cover, storm runoff paths, and distance from the sediment source should be noted to identify the conditions under which delivery may be appreciable. If these conditions suit the assumptions of one of the existing sediment delivery equations, then the equation might be applied. Results, however, should always be compared with field-based estimates. Wind erosion provides a special case of sediment delivery, because sediment from this source is not delivered preferentially to channels. Because the resulting input therefore is usually small, wind erosion often can be ignored in calculations of sediment production to streams. If needed, input rates can be estimated by assuming channel inputs are proportional to the fraction of the land surface occupied by channels and ponds.

3.7 Grain-size composition . Each erosion process produces a characteristic size distribution of sediment. Sheetwash erosion generally carries off only the finer sediment present on the soil surface, while shallow landslides remove the entire soil column and produce sedi- Chapter 3: Sediment production 51

Box 11 Why you can't describe the world from your desk -a true story- We have repeatedly stressed the importance of basing all analyses on a strong understanding of erosion and sediment transport processes, and, in particular, of selecting your methods according to what your field observations tell you is actually going on at the project site. This point can not be overemphasized, and we offer you the following true story in illustration. A hydrologist combined a computerized topographic data base with the USLE to calculate sediment production from a drainage basin with highly variable topography and land use in the mid-continental United States. The sediment yield from each subarea was then multiplied by a sediment delivery ratio based on the topographic characteristics of the source area. The results showed unequivocally that by far the largest sediment yield per unit area was in the undisturbed forests of a national park, whereas the cultivated and urbanized portions of the basin contributed relatively little. Skeptics were alerted, and a quick calculation showed that the predicted yields from the forest implied that the 80-year-old trees therein should be standing on large erosion pedestals, or at least show extensive exhumation of roots. An excursion to the area revealed not only that the pedestals did not exist, but that the high sediment yields predicted for the forest were simply the result of using an inappropriate model of the process of sediment transport (a sheetwash erosion equation was applied to a landscape devoid of extensive overland flow) and a sediment delivery equation that was exceedingly sensitive to the steepness of the basin topography. Fieldwork revealed no evidence that the forest was a significant sediment source in the area, and an early dose of fieldwork would have saved a good deal of confusion and embarrassment later on.

ment of the same size distribution as the soil profile. In the latter case, the contribution of each grain size to streams can be calculated if the grain-size distribution of the soil is known. Deep-seated landslides often contribute blocks of bedrock as well as colluvium, and the amount of bedrock eroded can be estimated by comparing the depth of the landslide scar to the average soil depth at similar sites. The maximum particle size transported to streams by sheetwash can be estimated from bar-like deposits in rills, road-side ditches, or low-gradient portions of the transport route. These grains represent the coarsest sizes in transport, and the deposits closest to the stream reflect the largest sizes contributed to the channel. Several workers have attempted to apply sediment transport equations to sheet and 52 Chapter 3: Sediment production rill flow. Moore and Burch (1986) suggest that stream power per unit weight of water is a good predictor of sheetwash transport, while Alonso et al. (1981) recom- mend use of the Yalin (1963) or Meyer-Peter and Muller (described by Graf 1971) equations on the basis of comparisons with laboratory results. These equations might be used to estimate transport rates for larger storms on the basis of samples taken during a single storm. Differences in process distribution, rates, and soil types cause differences in the sediment sizes contributed from different watersheds (figure 12). Reid et al. (1981) assess the grain-size distribution of sediment from various sources to calculate the influx of particular grain sizes harmful to fish habitat in coastal Washington. This type of analysis can also be used to predict changes in sediment character as changing land use alters the relative importance of erosion processes in a watershed .

.••..• TEAKETILE CATCHMENTS 80 ---- LONG MEADOW CATCHMENTS ... ../ -- RANCHERIA CATCHMENTS ...:: : :::::::~) :: WEST 60

o 0.002 0.008 0.031 0.125 0.5 2 8 32 128

GRAIN SIZE (MM)

Figure 12: Grain-size distribution of average annual sediment yields from 6 small (3.1 to 35.2 km 2) catchments in the N Fork Kings River basin, California, calculated on the basis of a sediment budget for each basin.

3.8 Calculation of long-term r~tes Many of the assessment methods described above provide rate estimates for particular time intervals. If climatic conditions in the interval were typical of the long-term record, then these estimates are likely to be good indicators of long-term rates. However, most sampling periods experience either a surfeit or deficit of high- magnitude events, and estimated rates must either be adjusted to reflect a characteristic event distribution or be used only to bound estimates of long-term averages. Kirkby (1987) notes that persistent periods of atypical event patterns are to be expected in the climatic and hydrologic record, so even long-term monitoring Chapter 3: Sediment production 53

records must be evaluated to determine how representative they are. Where the dependence of erosion rates on event magnitude is established, as is often possible for surveys of landslide frequencies (e.g. Reid 1989), rates are readily modified to represent long-term averages by summing values for a characteristic storm distribution. Reid (1989) also uses this approach to calculate long-term rates of plungepool erosion in gUllies. Where air photos spanning several decades are available and when the geomor- phologist has the time for a thorough analysis, the effects of exceptional storms or wet periods on landsliding can be analyzed by a frequency analysis. A simple example will illustrate the point. Suppose that the number of new landslides in a basin has been counted on successive aerial photos, and that it has been established through fieldwork and photo interpretation that there is a linear relationship between the number of landslides per unit area and some index of precipitation during the time interval between pairs of photographs (e.g. figur(! 13a). One could analyze the statistical properties of the long-term precipitation record for the region to determine the long-term frequency oflandsliding and, from this, the long-term frequency distribution of sediment influxes to channels from this source. As the air-photo record now approaches a length of 50 years in many areas, it is becoming possible to develop such relationships in a surprisingly large number of places. The simplest way to examine the problem is illustrated in Figure 13. The number of landslides per unit area generated in a 5-year interval by a maximum monthly rainfall of between 320 and 340 mm (53.4 per square kilometer in 5 years, estimated from the linear regression) is multiplied by the frequency of such an event (12.2 percent of all 5-year periods) to yield 6.5 slides per square kilometer in 5 years, or an avera~e of 1.3 slides per square kilometer per year. Such a calculation is repeated for other rainfall classes and the results summed to obtain the long-term total of landslides (figure J3b). Multiplication of the number by an average volume produces the long-term sediment contribution. Where the relation between slide frequency and the climatic parameter is linear and the climatic parameter is normally distributed, as in the case illustrated by Figure 13, the problem can be solved analytically using the approach described by Reid et al. (in preparation). First, the slide frequency (y) is regressed against the climatic parameter (x): y = a + bx + € (3.8)

where € is a normally distributed error term and y is defined to be zero for any values of a+bx+€ that are less than O. The mean (J..L) and standard deviation (cry) of the normal distribution that describes the slide frequency are then defined. Using a relation described by Ross (1988), the probability density of landslide frequency, !v(Y), can then be solved for:

2 '] ( y-I', J f() 1 -2 --;:- v y e ' (3.9) . = cry-ffi 54 Chapter 3: Sediment production

ro 150 16 2: Q) Landslides = 0.47 Rain - 102 A. 14 C r2=0.75 ooro ..... co e co 12 --2:'(ij Q) Rainfall Q) ..... >, 100 distribution I 10 .-cE:J LO ~ e Co ~ 0.>'-E N Landslide 8 >,X ICO E ~observations ~ LOE ..... 6 o e Q) 50 -...... Q) 0- e > 4 Q) '- 00 urn Q) ...... t:: :'Q 2 (jj &~ "0 e co 0 • 0 -l 150 200 250 300 350 400 450 500 550

Maximum monthly rainfall in 5-year interval (mm)

>, 7 u~e- Q) co r-r- :J 2: 6 B. - UQ) Q)+-' - ..... e 5 .- - -0.> "0 >-I = L() 4 00 - "0 .~ eN 3 r- .!!1E I.- 2 .....E~ en 0.> Q) +;'-0 rn= e en o~ 0 ~ -' ~ -, 150 200 250 300 350 400 450 500 550

Monthly rainfall responsible for landslides (mm)

Figure 13: Calculation oflong-term landsliding rate (from Reid and Dunne in preparation)

The long-term average landslide frequency is E(y), the expected value ofy, and is calculated as the integral of the product ofy andfv(y): t~llJ2 r E(y) = [ Yf.(y)dy = (j.e·. ~ + ----L11 { erf( 00) - erf [ - 11;;:; )}. (3.10) (j ., . " 21t 2 y ,,2 where erf represents the error function, which is graphed in Figure 14. Chapter 3: Sediment production 55

1.0

0.8

0.6

E :::l 0.4 0 0.2 -c: 0 U 0.0 c: ::J -0.2 L- 0 -L- L- -0.4 LU -0.6

-0.8

-1.0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0

Value of um

Figure 14: The error function (er!) ofUrn

For many applications it is useful to estimate the frequency of events larger than a particular size. For example, the occurrence of more than 90 landslides per square kilometer in the specified time interval might be associated with increased water treatment costs in an area. The frequency of such an occurrence can be calculated in two ways. First, the rainfall associated with a rate of 90 slides-km-2 during the interval can be calculated from equation 3.8 and the probability of achieving that or a higher maximum monthly rainfall estimated using Figure 13a. The frequency associated with the rainfall threshold equals the desired frequency for the threshold landsliding rate. The second method makes use of the analytical solution:

ply >n) = Haf(OO) -erf ( :~~)} (3.11 )

where p(y>n) is the probability that the number of landslides in an interval is greater than n. Of course, such an analysis is relevant only to an area where there has been no vegetation change, road building, or other disruption during the period of interest. If such a change has occurred, rarely is the change old enough that there is a long record of landsliding in response to the change, and one is limited to estimating the frequency of the triggering rainstorm and extrapolating to longer time intervals on that basis. 56 Chapter 3: Sediment production

Box 12 Calculation of long-term landslide frequencies For the case shown in Figure 13, Y = 0.47x - 102, so a=102 and b=0.47, where y is the landslide frequency for an interval with maximum monthly rainfall of x. The rainfall distribution of Figure 13a has a mean of rnx=303.5 and a standard deviation of sx=58.5, which are determined by examining the cumulative plot of the rainfall distribution. Because the maximum rainfalls in this case are normally distributed, Ilx is read as the value of x exceeded 50% of the time, and crx is the difference between this value and the value exceeded 68% of the time. The mean and standard deviation of the landslide frequency are then calculated as:

Il y == a + brnx = 40.6

cry ==~b2S; +s; =32.7 where s; is the mean squared error:

The long-term average frequency can then be calculated from equation 3.11:

E( )= 32.7e-C;4~}S + 40.6{erf (oo)-erf ( -40.6)} Y J2; 2 32.7J2 erf(oo) = 1.00 and erj{-0.878) = -0.785, so E(y) = 42.3 landslides in an average 5- year period, or 8.45 slides per square kilometer per year. The probability that no landslides will occur during a 5-year period is calculated as the difference between 1.00 and the probability that the number of landslides is greater than zero (from equation 3.12):

p(y=0)=1.0-p(y>0)=1.0--1 { erf(oo)-erf ( -40.6)}J2 =0.108 2 32.7 2 For zero landslides to occur during a 5-year interval, zero landslides must have occurred during each of the 5 years (zero in year 1 and zero in year 2 and... and zero in year 5), so the probability of no landslides occurring in a year is:

1 P(YI = 0) = P(Y5 = 0)5 = (0.108)5 = 0.64 (continued) Chapter 3: Sediment production 57

continued: Box 12 The probability that more than 90 slides will occur in a 5-year interval can be calculated directly from equation 3.11:

p(y > 90) = -1 { erf ( 00) - erf (90 - 40.6)}J2 = 0.066 2 32~ 2 Because a total of 90 or more slides in. a 5-year period can be achieved by many different combinations of landslides in each of the 5 years, there is no easy way of calculating the probability that more than 90 slides will occur during a single year.

Less ideal situations may arise when only two or three sequences of photographs are available and a statistical analysis is not possible. In such a case, one might use an average rate from the available record and define the nature and potential magnitude of deviations from that average. For example, let us suppose that only two aerial-photo sequences are available and they include the effects of landslides triggered by a 100-year event. Application of the measured landslide contribution rate to a longer time period would overestimate the long-term sediment supply, but would serve as an upper bound on that rate. However, the data also provide a useful definition of potential short-term influx rates, and for some purposes these may be more important to define than long-term averages. In addition, the photos would provide some information about the duration and magnitude of sediment inputs during recovery from such an event, since the date of the event would be known. If no landsliding has occurred during a short aerial-photo record, there is an even greater difficulty. Should the absence of landsliding be attributed to the inherent stability of the terrain, or to the lack of a sufficiently large rainstorm on favorable antecedent conditions? The only way to discriminate between these hypotheses is to search for indicators of past mass wasting, such as patches of young or even-aged vegetation, landslide scars and deposits on hillslopes, or channels that have been scoured or aggraded by landslide-triggered debris flows. Some of these features, such as the age-distribution of riparian tree canopies, may be interpreted from aerial photographs, but recognition of subtle depositional evidence usually requires fieldwork. In any case, field checking will be necessary for confirmation and for dating the evidence. Dating is usually possible by measuring the ages of recoloniz- ing plants. Once an age has been obtained for a pulse of landsliding, a check of weather records may yield an estimate of the size of the triggering storm or of the annual or seasonal rainfall total. For some purposes, knowledge ofthe approximate frequency of such pulses might be adequate. For other projects, one might need to map as much evidence as possible of such an event. In the latter case, mapping of all 58 Chapter 3: Sediment production datable scars or deposits would produce a sediment influx which could be combined with the estimated storm frequency and (with the caveats outlined above) incorporated into a full sediment budget. If such an analysis is undertaken, it is essential to decide what level of precision is required for the result to be useful. A full analysis clearly requires extensive fieldwork, but most problems can be resolved by very approximate estimates oflandslide frequencies for which only a partial evaluation is required. Where land-use changes have b~en recent, all possible measurements may represent only a fraction of the progress of an erosional cycle brought about by the land-use activities, and it is necessary to define the change in rates through time if long-term averages must be estimated. Surfaces newly cleared of vegetation often experience rapid erosion during the first few seasons but recover rapidly. Megahan (1974), for example, measured exponential decreases in erosion rates from such sites. Similarly, rates of road~related landsliding often show a rapid decrease with time after road construction (figure 15), and landsliding rates measured early during the intrusion of land management into a basin will overestimate their long-term effect. To account for this change, landsliding rates can be defined as a function of road age by measuring landslide age and incidence on roads of a variety of ages. Aerial photographs and road maintenance information are usually adequate for this in areas managed for timber production.

0.7

.....-- 0.6 -0-- debris avalanches >.E --T- debris flows u~ c"o O)ro 0.5 :::JO 0- .... 0) .... 0.4 '-+--a..... 0) 0) .... :Qro 0.3 -0) ~>. c .... 0.2 roO) -Ie.. ~ 0.1

0.0 0 2 4 6 8 10 12 14 Years after road was built

Figure 15: Landslide frequency versus road age for road-relatedfailures for sites in the Clearwater basin, Olympic Peninsula, Washington. Landslide ages were determined by presence or absence on sequential aerial photographs, so frequencies represent running averages over multiple-year periods. Chapter 3: Sediment production 59

3.9 Predicting future erosion rates Sediment budgeters are often asked to predict the likely impact of proposed land- use changes. Such predictions are possible if the factors that control process rates are known and if the nature of that control is understood. In many cases the relation is quite simple: one can easily calculate an expected landslide frequency if a landsliding rate is reported as a frequency per unit length of road and if plans specify the length of road to be built across similar terrain in the future. Stratification criteria must be selected carefully when subdividing the project area to allow this extrapo- lation, and process rates should be rel,,!ted to causal factors wherever possible. Where causal factors are not well understood, multiple influences can be cautiously combined by defining process rates per unit area of each type ofJand use. Future land-use activities often will be different from those already present in a project area. Practices may be improved to lessen impact; or less favorable areas may be entered, increasing impact; or an entirely new activity may be introduced. Qualitative and order-of-magnitude predictions of sediment inputs are usually possible in these cases by observing the effect of the anticipated activities in areas where the change has already occurred. The effect of new practices can also be predicted by evaluating their influence on the variables that control a process rate. Paving of forest roads, for example, removes road surfaces as a source of sediment, but may increase peak discharges in road-side ditches and low-order channels. The approximate magnitudes of these effects can be calculated or estimated from observations at analogous sites, and the order of the expected change can then be used to infer the effects on sediment input to channels. Similarly, landslide frequency is often highest soon after roads are constructed, so the proportional sediment input from road-related landsliding will decrease after a road network is completed, but rates of surface erosion will remain high. Unfortunately, future land use is rarely completely described by plans. Most changes provoke other, unintended changes. Thus, construction of roads in a watershed usually increases recreational use or agricultural settlement, and the effects of these secondary activities must also be evaluated. This analysis requires prediction of both the nature of the indirect effects and the impacts resulting from them. Such evaluations usually are qualitative and are based on field examination of sites at which similar developments have already occurred. Predictions are also uncertain because some geomorphological processes react to changes in obscure ways. In some cases, altered process rates may not be evident until a rare, high-magnitude event occurs. Road culverts are often designed to pass a 25-yr recurrence interval flow, for example, so the eventual impacts of this practice will not be evident until a larger flow occurs. Similar problems plague analyses of slope stability. The heavily logged Redwood Creek basin experienced severe landsliding and aggradation triggered by heavy rains in 1964. With only the single storm's legacy and no large, unlogged basins for comparison, it would have been 60 Chapter 4: Channel transport and storage difficult to detennine how much of the damage was caused by land use and how much would have occurred anyway during such a large stonn. However, a similar stonn in the late 1800s that predated logging apparently had little effect on the basin (Harden et al. 1978). Logging practices have changed since '1964, and the magni tude of watershed damage has decreased. Is the decreased severity of impacts due to the altered practices, or does it simply reflect the absence of major stonns? Where the expression of impacts depends on the occurrence of major events, the reliability of predictions decreases. Such effects usually can be assessed only qualitatively.

4 EVALUATING SEDIMENT TRANSPORT AND STORAGE IN CHANNELS

Although the evaluation of sediment input to channels is sufficient for many sediment budget applications, others require an understanding of what happeris to sediment after it enters a stream. These applications include assessment of the effects of altered sediment or flow inputs on channels (table 6). The effects of most concern generally are changes in water quality and changes in channel or reservoir morphology brought about by erosion and deposition. To evaluate these effects, it is necessary to understand: how much of each type of sediment is transported and where it is deposited; how that sediment interacts with sediment in storage; and how the transport affects channel morphology. Analysis of potential changes usually requires both characterization of a channel under undisturbed conditions and prediction of how those characteristics will change with altered inputs. In some cases, answers to these questions are very complex. Sediment transport can provoke cumulative effects at sites far removed from the original sediment inputs; changes in transport rates can undennine channel banks and generate new sources of sediment; a change in sediment input itself alters transport rates; particles of different sizes travel at different average speeds; and the particles in transport are broken down by abrasion and weathering. Channel-related questions cah often be answered only qualitatively because of such complications, but the ability to make quantitative predictions is improving. The fate of sediment in channels depends on the volumes and grain sizes intro duced, their mechanical and chemical durability, the sediment-transport capacity and competence of the stream, and the opportunities for detention along a river. The capacity of a stream for transport of a particular grain size is defined as the quantity of that grain size which the stream can transport past a cross section in a unit of time (e.g. tonnes per second or per year). The competence of a stream refers to the maximum size of mobile sediment grains. Chapter 4: Channel transport and storage 61

Table 6 - Examples of questions concerning channels that can be addressed by sediment budgeting, and a selection ofanalytical techniques that can be used to answer them. Analysis required*: Channel-related problems C R F G M T A H Effects of changes in sediment input I. How much introduced sediment will be carried away? G M A H 2. Where will the introduced sediment be deposited? C G T 3. How much deposition will occur? C F G T 4. What grain sizes will be deposited? G M 5. How will increased sediment affect channel form? G M T A H 6. How will reduced sediment affect channel form? G M T 7. Will this site be stable if sediment input changes? G M 8. How long will it take the channel to recover from a briefly F T altered sediment input? 9. How will altered sediment input affect water quality? G T Effects of changes in flow I. Will a flow change cause incision? F G M 2. Where will incision occur? C A 3. Will a flow change cause aggradation? F G M 4. Where will aggradation occur? C G T A 5. How will channel form change to accommodate altered flow? G T H 6. How will the altered flow affect scour depths? C R F G M Specific applications I. How fast will a reservoir lose storage capacity? G T A 2. What is the expected sediment yield in a watershed with a C R G T A H particular distribution of land uses? 3. How long will it take a channel to recover from a particular C R F G M T A H impact?

* Method: C Channel characterization (slope, form, G Grain-size analyses downstream change, etc.) M Mobility calculations RUse of records or evidence from similar T Sediment transport calculations channels A Analysis of sequential aerial photographs F Definition of flow-duration curve H Evaluation of historical evidence

Despite the wide variety of channel-related questions that commonly confront geomorphologists, the types of analyses required are relatively few (table 6). These include field measurements of channel characteristics (e.g. slope and channel dimensions), use of records from similar channels, definition of flow-duration curves, measurement of grain-size distributions, determination of mobile grain sizes, estimation of sediment transport rates, evaluation of historical and vegetation evidence for channel changes, and analysis of sequential aerial photographs. This chapter describes the application of these techniques both to characterize existing channel conditions and to predict likely changes. 62 Chapter 4: Channel transport and storage

4.1 Overview of channel processes The character of sediment production and transport changes predictably along a channel system, as do the components of the sediment load and their interrelation ships (figure 16). Colluvium is the loose, weathered material that mantles hillslopes and is moved downslope by processes considered in the previous chapter. Some of the colluvium that enters a small, headwater channel is too large for the stream to transport. This sediment accumulates until it either breaks down into more trans portable sizes or is scoured away by debris flows, which can mobilize the entire channel bed as a slurry.

BED MATERIAL

BED MATERIAL SUSPENDED LOAD

Figure 16: Components o/the sediment load in a channel

However, most of the contributed colluvium is small enough to be transported at least by occasional storm flows. Sediment fine enough to be flushed downstream in suspension without settling is called the washload, and does not accumulate in significant quantities in channel deposits except where it is stranded on floodplains by overbank flow. Washload is a large fraction of the suspended load in many streams, but it is not readily predicted by equations because its transport rate depends more on the rate of supply from hillslopes than on channel hydraulics. At present, the only way to document its short-term magnitude is to measure it with a suspended-sediment sampler and relate the flux rate or concentration to some easily available streamflow parameter, such as discharge. A recent study suggests that relationships between washload concentration and discharge are strongest when only the quickflow component of a hydro graph is considered (Mizumara 1989), and Dunne and Leopold (1978, p. 287) outline methods for separating hydrograph flow components. Long-term rates of washload production can be estimated from a hillslope sediment budget by comparing the grain sizes entering the channel to those present in the channel bed; the components missing from the bed constitute the washload. Chapter 4: Channel transport and storage 63

The coarser fraction of the incoming sediment settles to the bed at least intermit tently during its downstream migration and so forms the bed material. A portion of this bed material is suspendible and is known as the bed-material suspended load (figure 16); the remainder travels along the stream bed as bedload. A particular grain size may travel through a reach as bedload at one stream discharge and in suspension at higher flow, and the mode of transport may also change as channel morphology varies downstream. For most gravel-bedded streams, however, the distinction is easy to make for the dominant transporting flows. When using these definitions to understand sediment transport, it is important to keep in mind that the operational definition of bedload is "that which is caught in a bedload sampler." Because the current generation of portable samplers extend 7.5 or 15 cm above the bed, they also trap the coarser, most concentrated portion of the suspended load. Similarly, the operational definition of suspended load is "that which is caught in a suspended-Ioed ampler", and this includes both washload and bed-material sus pended load traveling more than 7.5 cm above the bed. Sediment changes character during transport by breaking apart or being rounded by abrasion, while chemical weathering continues during periods of storage in channels and on floodplains. Sediment in high-order, lowland channels has a finer grain-size distribution than that of low-order, headwater channels because most coarse particles have been rounded and weak clasts have broken down. Lowland streams are usually alluvial; they flow through channels they have molded from their own sediment, and they rarely encounter the colluvium or bedrock of valley walls. Colluvial transport on lowland valley walls usually contributes sediment only to low-order tributaries and to the surfaces of floodplains and terraces.

4.2 Characterization of channels If the applications intended for a sediment budget require that channel processes be evaluated, measurements usually must be made of some channel characteristics. These measurements may be used to calculate sediment transport in channels or simply to describe the channel to allow comparisons with other channels or recog nition of future changes. Channels are usually characterized by their slope, cross sectional geometry, roughness, flow characteristics, rates of channel change, and past channel changes.

4.2.1 Qualitative characterization A cOf\Siderable amount of useful information can be gained simply through qualitative observations of channel character, and this is usually a preliminary step necessary for determining what other analyses are needed and where to make . measurements. Much of this work can be done on aerial photographs, and most is directed toward identifying the processes and influences that are most important in determining the channel character. 64 Chapter 4: Channel transport and storage

A quick perusal of aerial photographs will usually allow division of a channel into distinctive reaches according to whether they are bounded by floodplains and terraces, and thus contain potential sedimeng snorage sites, or are constrained by bedrock and colluvial slopes, and so are more likely to serve primarily as transport reaches. This categorization provides an immediate guide to the locales most useful for particular kinds of measurements or for diagnosis of past channel changes. Alluvial reaches are at greatest risk of aggradation, and aggrading'sites can often be recognized on aerial photographs because of their vegetation characteristics and surface morphology. Undissected alluvial fans, for example, usually imply a long history of aggradation. Aerial photographs and field observations generally reveal the sediment sources that most strongly affect channel processes. Debris flows and near-stream landslides are particularly important and should be noted. Landslides provide point sources of coarse sediment that strongly influence channel morphology and the size of sedi ment available for transport, and debris flows create characteristic channel forms and affect sediment input to downstream reaches. If such processes are recognized, sites for detailed measurements can be selected to avoid their effects. Some · areas may have controlling influences that are less common, and these must also be understood if sediment budgeting measurements are to be correctly interpreted. Watersheds containing glaciers or karst-controlled drainages, for example, cannot be evaluated without taking these factors into account. Many forested channels are strongly affected by fallen logs, and field reconnaissance will disclose the channel sizes for which these are an important influence. These types of influences on channel form and pattern must be identified in the field before more detailed measurements are made. Most sediment budget applications in channel systems require diagnosis of the types of flow events that transport significant amounts of sediment and determine the channel character. High flows are usually generated either by rain, snow melt, or rain falling onto a snow pack, and rain-on-snow has consistently produced the most severe floods in northern California and the Pacific Northwest (Harr 1981). The relative importance of these modes of runoff generation is usually evident from climatic records and descriptions from residents. By the end of the preliminary characterization, one should have a good qualita tive, but systematic, understanding of the major sediment sources for the channel, their location, and the grain sizes contributed. The major factors that control runoff and channel form should also be identified, as well as their variation through the watershed; and the distribution pattern of alluvial reaches, depositional zones, and areas of strong hillslope influence should be known. Chapter 4: Channel transport and storage 65

4.2.2 Selection of measurement sites The success and utility of channel measurements usually depend strongly on the site selected for measurements. Natural channels are so complex that it is easy to find sites at which measurements would be meaningless. Conversely, it requires understanding of channel processes and intended measurement applications to select appropriate measurement locations. Channel form varies at many scales. Over a distance of tens of kilometers, channel gradient usually decreases in a downstream direction, channel dimensions increase, and channels become progressively less influenced by local hillslope processes. Over shorter distances of tens to hundreds of meters, channels usually show rhythmic sequences of bends, gravel bars, and alternating low-gradient pools and high-gradient riffles. Over even shorter distances, channels are perturbed by fallen logs, bedrock outcrops, and other obstructions. All of these types of variation must be recognized if a suitable measurement site is to be selected. Channel char acter also varies markedly through time as discharge changes or obstructions form or disappear. Measurements are generally made for one of three reasons, and each requires a particular type of site. Measurements to be used in calculations of flow or sediment transport should be taken at sites free of local complications. Channels should be single-stranded and unobstructed, and the reach selected should be as straight and flume-like as possible to allow sufficient development of uniform flow. Ideally, the reach should be alluvial rather than bedrock-bounded, and the bed surface should be of homogeneous texture, rather than being composed of patches with contrasting grain sizes. In some cases, channel characteristics during the high flows relevant to the calculations are quite different from those present at the time of measurement. A low-flow, single-thread channel may become part of an anastomosing network at higher flows, for example. In such cases, the high-flow characteristics· must be reconstructed as accurately as possible if calculations are to be useful. Where measurements are to be compared to those from other streams, the sites to be compared must be equivalent, and the type of site to be measured should be selected according to the reason for comparison. For example, if comparisons are to be made to test for channel widening in a logged basin, then an appropriate site-type might be straight, alluvial reaches. It would be useless to compare such measure ments with widths at bedrock-bound sites, and a random comparison that did not account for inherent variations in width between pools, bends, riffles, and straight reaches would require a prohibitively large sample size. Measurements are often made to provide a baseline for detecting future changes. In this case, the likely nature of the change must be predicted, and sites must be selected at which such a change will be visible. If aggradation is expected, for example, the most sensitive sites will be pools in alluvial reaches or sites currently 66 Chapter 4: Channel transport and storage

Box 13 Selecting a useful channel reach for analysis Characterizing channel geometry: 1. The measurement reach is straight and contains no obstacles to flow 2. The channel is bounded by alluvial sediments and is self-formed 3. Flow is in a single thread for the discharge range of interest (if the channel is braided, select as simple a reach as possible) 4. Water-surface gradient is uniform through the reach 5. The reach is not near a tributary confluence Characterizing channel-bed sediments: 1. The measurement reach is straight 2. Alluvial sediments are present in bars or on the channel bed 3. Grain sizes are homogeneously distributed through the reach 4. Sediment-transporting flows are in a single thread (if the channel is braided, select as simple a reach as possible) 5. The reach is not near a tributary confluence or major sediment source unless this effect is to be evaluated Calculating sediment transport rates: 1. The measurement reach is straight and contains no obstacles to flow 2. The channel is bounded by alluvial sediments and is self-fomled 3. Water-surface gradient is uniform through the reach 4. Grain sizes are homogeneously distributed through the reach 5. Sediment-transporting flows are in a single thread (if the channel is braided, select as simple a reach as possible) 6. The reach is not near a tributary confluence or major sediment source Assessing recent channel changes (i.e. sites at which changes are likely): 1. The reach is bounded by alluvial sediments 2. Channel margins are conducive to riparian growth 3. The reach contains bends and complex channels 4. Gradient through the reach is not uniform Reconstructing a particular discharge: If a reach receives no tributary inputs or distributary losses, and is of relatively uniform morphology, peak discharge will be relatively uniform along the reach. This means that an indirect estimate of discharge anywhere along the reach will be appropriatt!; you can estimate it wherever the evidence is clearest. In particular, methods exist for estimating discharge at flow constrictions (Matthai 1967), cul verts (Bodhaine 1968), and overfalls (Hulsing 1967). Though these are not conven ient sites for sediment transport calculations, the discharges estimated can be used for transport calculations elsewhere along the reach as long as these features do not affect the sediment supply downstream. Chapter 4: Channel transport and storage 67

undergoing deposition. Sites that are susceptible to local perturbations should be avoided. Measurements made near a tributary confluence, for example, may more closely reflect changes in that tributary than in the rest of the drainage basin, and sites near a debris jam will eventually be complicated by break-up of the jam.

4.2.3 Slope measurements Channel slope is a relatively conservative channel parameter because a measur able change in channel gradient requires relocation of a large amount of sediment. Slope is thus a worthless parameter for monitoring, but it is very important as a controlling variable for all flow and sediment-transport calculations. The slope which is used in such calculations is that of the water surface, and this will vary as a function of both location along a reach (e.g. between pools and riffles) and dis charge at a site (differences in water-surface gradient between pools and riffles usually decrease as discharge increases). If slope is to be used in flow or transport calculations, the necessary slope is that present during the discharge of interest rather than that present at the time of measurement. This is awkward for two reasons: the discharge of interest must be known, and the relevant slope must be reconstructed. If a characteristic transport rate is required, then the reference discharge usually is selected as the bankfull discharge, both because this is definable in the field and because it is considered to be a dominant discharge that controls the dimensions of alluvial channels. This morphologically-defined discharge also provides a useful reference discharge for comparisons between channels. The water-surface slope during bankfull discharge can be approximated by surveying the margin of the floodplain along a reach for which the calculations are to be made. On low-gradient channels, water-surface slopes generally are uniform along the channel at bankfull stage, and the gradient can be calculated as an average along a relatively long measurement reach. Note, however, that the downstream distance used in calculat ing gradient is that of the flow at bankfull discharge, rather than the length of the low-flow thalweg between elevation stations (figure 17). A hand-level, rod, and tape usually provide enough precision for flow and transport calculations on high gradient channels, but the distances and small angles that must be measured on low gradient channels usually require use of a surveying level. Slopes are occasionally measured to permit transport or flow calculations for a particular flood, and definition of slope in these cases is usually easier. The height of the peak stage is evident from deposits of rafted debris or removal of floatable litter for a short time after a high flow. These lines are more easily recognized than bankfull heights, and are readily surveyed. Identification of crest heights at several points along a reach allows measurement of water-sU{face gradient. 68 Chapter 4: Channel transport and storage

Box 14 Recognizing bankfull stage Site selection: I. Use an alluvial reach, where the channel is self-fonned. 2. Avoid locations showing evidence of bank erosion. 3. The presence of well-defined floodplains makes the task easier. 4. Bankfull discharge changes little through unifonn reaches without tributary inflow, so well-defined values from a nearby site can be more useful than poorly defined infonnation at the particular site you are interested in. 5. Select several sites along the reach to check your estimates. What to look/or: I. A topographic break from the steeply sloping bank to the subdued floodplain topography 2. Changes in vegetation from bare to grass, moss to grass, grass to sedge, or treeless to trees 3. Change of texture and bedfonns in deposited sediment 4. Highest elevation below which there is no fine organic debris 5. Sharp change in the evidence of sediment transport intensity (e.g. wash marks, bedfonns, and oriented and imbricated particles)

Examples of surveyed cross sections showing bankfull stages are illustrated by Dunne and Leopold (1978, pp. 597-606) and by Leopold (1974, pp. 77-99).

4.2.4 Channel geometry measurements Measurements of channel width and depth are the most common basis for inter channel and temporal comparisons, and they are required also for transport and flow calculations. The usual measure for comparison is the bankfull width and depth, since bankfull discharges generally have similar recurrence intervals in channels of similar orders throughout an area. As with channel gradient, width and depth usually vary predictably from reach to reach, and a standard site type must be selected if valid comparisons are to be made. Straight, alluvial reaches without introduced obstructions again provide the most useful sites for comparison. Measurement methods are comprehensively described by Benson and Dalrymple (1967). Chapter 4: Channel transport and storage 69

High-flow channel ---__

Measurement points B' o 10 m

Low-flow water-surface B profile

__.....::B'~ o '------'------'------"""datum o 50 100 150

High-flow water-surface profile

'B ,-datum O'------'-----~~-~ o 50 100 m Distance along thalweg from A - A' (m)

Figure 17: Example ofthe relation between low-flow and high-flow water-surface profiles

4.2.5 Channel roughness Calculation of discharge from channel geometry measurements requires an estimate of the roughness of flow in the channel. If roughness is known, discharge may be estimated as a function of stage. Roughness in alluvial channels without major flow obstructions is usually described by Manning's roughness parameter, n. For large channels, Manning's n is most easily determined by comparing the chan nel to the calibrated photographs presented by Barnes (1967). Values of Manning's n range between about 0.01 in smooth, sand-bedded channels, to higher than 0.06 in steep, rocky channels. Values can also vary considerably as discharge increases at a site and progressively immerses cobbles arid boulders, but roughness usually varies 70 Chapter 4: Channel transport and storage only slightly near bankfull stage. Back-calculation of Manning's n from current meter records at USGS gauging stations on similar channels often provides the best estimate of roughness for a particular channel. Where Manning's n can be estimated, the Manning equation can be used to calculate discharge or velocity (see later, equation 4.2). Manning's equation is less useful for steep channels with large obstructions. Thome and Zevenbergen (1985) compared measured velocities in a steep creek with those calculated using five roughness. equations and found that agreement was generally poor. The best predictor of average velocity (u, mls) was found to be the equation developed by Hey (1979):

u = 5.62J gRS log ( --a'R ) (4.1a) 3.5Ds4

-0.314 a' = 11.1 ( ~ ) (4.1 b) hmax

where g is acceleration due to gravity (mM), R is the hydraulic radius (m), D8./ is the

particle size coarser than 84% of the sediment (by volume), and hlllax (m) is the maximum depth at the section.

4.2.6 Definition of flow characteristics Methods for measuring discharge and velocity are described in many texts, and standard techniques have been developed by the U.S. Geological Survey and other agencies (Herschy 1985). Geomorphologists often must reconstruct the discharge of a particular flow event from evidence of flow depths, and most sediment transport calculations require estimates of flow-duration curves. Reconstruction of discharge requires estimates of flow depth, cross-sectional area, water-surface slope, and channel roughness. Reconstruction of a particular discharge is easiest on low-gradient alluvial reaches where roughness can be described by Manning's n. Discharge (Q, m3/s) can then be calculated using the Manning equation:

Q = A R2/3 S1!2 (4.2) n where R is. the hydraulic radius (area/perimeter, m), S is the water-surface gradient, and A is the cross-sectional area (m2). In rough channels, velocity can be calculated using Hey's equation (Hey 1979; Equation 4.1 above) and multiplied by the recon structed cross-sectional area to estimate discharge at a particular stage. Williams (1978) uses measurements of bankful discharge in 233 channels ranging from 0.7 to 8510 m2 in cross-sectional area to define an average relationship Chapter 4: Channel transport and storage 71 for estimating bankfull discharge (Qb' m3/s);

Qh = 4.0A; 21 S028 (4.3) where Ab is the bankfull cross-sectional area (m2) and S i;S the water-surface slope. Riggs (1976) outlines an equivalent method for reconstructing discharge (Q, m3/s) at flows less than bankfull in simple alluvial channels using the cross-sectional area of flow (A, m 2);

Q = 3.394 A 1.30 S032 (4.4) Each equation will work well for certain types of channels and not for others, so each must be tested against measurements to be sure that it is appropriate for a particular channel. Benson and Dalrymple (1967) describe other procedures for reconstructing discharges. Unless the channel of interest is gauged, flow-duration curves must be approxi mated by those measured for similar channels in the region and suitably corrected for differences in drainage area. The frequency of particularly high peak flows is often of concern in sediment budgeting, and historical evidence is often useful for providing this information. Residents can usually point out approximate water lines for several past floods, and some sites may preserve datable deposits from multiple overbank floods. Riparian vegetation is often scarred by floods, and Sigafoos (1964) describes the reconstruction of flood history using evidence from vegetation.

4.2.7 Identification of channel changes Changes in channel morphology reflect the erosion, transport, and deposition of sediment, and methods for determining these rates are discussed more fully in a later section. However, such analysis is also relevant to questions of expected channel stability and past channel behavior, and a description of channel behavior is an integral part of most channel characterizations. In particular, it is useful to know if parts of a channel system are aggrading or incising or have done so in the past, and if rates of channel migration have changed. These analyses typically require the use of historical information. The most useful source of historical information is often sequential aerial photo graphs, which provide accurate pictures of the landscape at intervals that extend back to the 1940s in most areas. Comparison of sequential photos can reveal the types of channel migration and morphological change that different parts of the channel network have undergone (figure 3). Early topographical maps can often extend the record to the late 1800s, though with less resolution, and notes from the original land surveys can occasionally provide an even longer view of the channel (Galatowitsch 1990). Channels can be transferred from maps or photographs at different scales to a single base map with the aid of a zoom-transfer scope, and images re-scaled using a variable-reduction copy mac'hine have proved adequate for 72 Chapter 4: Channel transport and storage

Box 15 'Considerations in the use of historical evidence Aerial photographs: these are generally the most useful source of information 1. It is easiest to compare photos of comparable scale. 2. Photographs taken during equivalent seasons are most useful. 3. Be careful to ensure that apparent geomorphological changes are not simply changes in vegetation or shadow lengths. Topographic maps 1. There is no standard for which streams are colored in blue on USGS maps, and definitions may have changed between map series. Changes in drainage density are thus difficult to extract from maps, while changes in channel pattern can be more precisely determined. 2. University libraries often retain old map editions in their collections. Anecdotal evidence 1. Extremes are often remembered more readily than average conditions. 2. Examine the field site with the informant. 3. Ask for details rather than generalities; in trying to evaluate aggradation, for example, you might ask "Which rocks did you used to dive from?", or "Where was the summer-time ford?". Similarly, ask to be shown the level to which the flood rose, or the location of the gully when the barn was built. 4. Poorly defined dates can often be defined more precisely by relating them to particular events that have little connection with the feature you are interested in. Questions like "What were you doing when you first noticed it?" can often re awaken associations that lead to more precise information. S. Some observers will have more reason to remember particular events than others. Whitewater paddlers can describe subtle changes in rapids, while anglers will know more about aggradation in pools. 6. Most people will unconsciously tend to tell you what they think you want to hear, so be very careful how you word your questions. You may want to bury the most important questions among others that are not as critical. 7. Ask to be shown the dimensions, rather than depending on a respondent's esti mate of a size. "Could you jump across it?" usually provides more accurate in formation than "How wide was the creek?" 8. Do not expect impartial answers to questions that relate to issues of local economic or political importance. 9. Always seek confirmatory evidence from other sources. Photographs 1. Check archives of local historical societies, and ask long-term residents. 2. University libraries often have collections of historical photographs. (continued) Chapter 4: Channel transport and storage 73

continued: Box 15 3. Copy the photographs and try to locate the camera location and view in the field; examining the site can produce a wealth of auxiliary information.

Land survey notes 1. These are usually held in state archives or those of a designated university. 2. Quality varies a lot between surveyors, and some records were faked, so always check the records against unchanged landmarks in the field. 3. Many notes will be cryptic, hard to read, or in abbreviations; read further through the notes to identify unknowns by context in later passages.

less precise comparisons. Local residents can often provide accounts and snapshots that offer clues to channel behavior. Buildings and other structures may also indi cate modes and rates of change. For example, Reid (1982) documents extreme channel widening at a site in northern Tanzania by comparing present channel width with that indicated by the spacing of ruined bridge abutments. Analysis of riparian vegetation can often extend evaluations of channel activity beyond the written record. Rates of aggradation or incision can also be estimated from riparian structures in some places. Haible (1980) estimated aggradation rates by measuring the depth to which datable fence posts were buried. Collins and Dunne (1989) determined the magnitude of recent incision by re-surveying channel cross sections at bridge sites that had been surveyed decades earlier. They also estimated rates of incision by plotting through time the water-surface elevation for a fixed low discharge near the average annual minimum. Upward or downward trends in the lowest water-surface of each year indicate raising or lowering of the bed. Necessary data can be obtained from summaries of current-meter gauging conducted by monitoring agencies such as the U.S. Geological Survey. Collins and Dunne were able to relate such bed elevation changes quantitatively to perturbations of the channel sediment budget caused by gravel mining in the channel. Several studies have compared channel geometry in a project channel to geome tries of analogous undisturbed channels to document the morphological changes a channel has experienced (e.g. Madej 1982). This method has also been used to predict the effects of anticipated changes in land use (e.g. Rango 1970, Jackson and Van Haveren 1987).

4.3 Grain-size distributions Both qualitative and quantitative analyses of sediment transport require informa tion about the grain sizes available for transport, so it is useful to begin an analysis 74 Chapter 4: Channel transport and storage of sediment transport by documenting the grain sizes of sediment entering the channel system, as described in Chapter 3, and comparing them with the grain-size distribution of bed material. Such a comparison allows estimation of the proportion of the input that is flushed rapidly downstream as washload, the proportion that is large enough to accumulate for long periods without transport, and the proportion represented by the mobile bed material. For example, Benda and Dunne (1987) show that the bed-material textures in channels of low-order streams in an Oregon watershed are similar to the grain-size distribution of soil on neighboring hillslopes (figure 18), and infer that these streams are capable of selectively removing sedi ment only from a thin surface layer of the colluvium deposited in the channels. Instead, most colluvium is flushed from these first- and second-order channels by episodic debris flows.

::~J-__ ~~__ -L__ ~ __~==~-==H=O=LL=O~W 1ST - ORDER

2ND - ORDER

3 ~ 4TH -ORDER ::1~ __ ~~__ ~ __ ~ ______~~~~ B C G P S FS S&C

Figure 18: Grain-size distribution of colluvium in hollows and alluvium from channels in a 4th order basin in the Oregon Coast Range. B = boulders (>512mm). G = gravels (64-22mm). P = pebbles (22-lmm), S = sands (I-0.25mm), FS = fine sands (0. 25-0. 06mm}, S&C = silt and clay «0.06mm) Source: Benda and Dunne (I 987}.

In documenting the grain size of bed material, the first complication to take note of is tha~ in many gravel-bedded streams a coarse pavement or armor layer covers a finer substrate (figure 19). The term "pavement" is used if the coarser layer is periodically mobilized, whereas "armor" is used if it is immobile and has resulted from the winnowing of mobile particles after a change in sediment supply or transport conditions. Because the distinction between the two surfaces depends to Chapter 4: Channel transport and storage 75

_ subsurface 25 _ surface ...... ffi 20 ~ Q) c.. 15 ...... c a> ~ 10

5 o 2.8 4 5.6 8 11 16 22 32 45 64 90 128 Grain size (mm)

Figure 19: Grain-size distributions of channel sediment at River Mile 22 of the Snoqualmie River, WA .

some extent on flow conditions, the type of surface generally cannot be recognized simply on the basis of sediment texture. Instead, identification is usually based on: evidence of recent changes in sediment or flow conditions (e.g. armoring is likely below a stream-side rockfall, or where flows have been diverted); evidence of bed stability (e.g. well-vegetated, coarse-surfaced gravel bars or tightly packed, algae covered, bouldery beds imply infrequent mobilization); and calculations of particle mobility (see following discussion). There is usually an approximate relation between the mean particle sizes of the surfl;lce and subsurface, but it is rarely strong enough for reliable prediction of subsurface texture from that of the surface. If the bed is armored or paved, the Box 16 surface layer may need to be sampled Evidence of channel-bed stability separately or removed before samp ling. This decision depends on the Gravel bars are vegetated Large clasts are angular particular sedimentation problem of Large clasts are densely packed interest. The texture of the coarse laye_ Bed is covered by large clasts is crucial if the concern is the magni Moss grows on bed sediment tude of flushing flows required for disrupting the pavement, while if the texture of the high-flow bed-material load is being studied, then the surface layer must be removed before the substrate is analyzed. This decision also affects the technology used in sampling. For 76 Chapter 4: Channel transport and storage

Box 17 A rapid field method for sieving gravel Obtaining reliable subsurface grain-size data for coarse-grained channel sediment is made difficult by the large sample size required to accurately represent the sediment population. The following field sieving technique (Booth et al. 1991) takes advantage of the fact that only the largest clasts require a huge volume of sediment for characterization, and so, conversely, only a fraction of a large sample is needed to characterize the distribution of smaller clasts.

Sample extraction proceeds as follows. First, a 1.5- to 2-m2 area is stripped of its pavement to a depth of 1 to 2 grain diameters, and a 1-m by 1-m frame is laid within it to define the sampling area so that the subsurface sample is not contami nated by clasts from the pavement. The gravel is then loosened with a pick-axe to a depth of a few tens of centimeters, and the sediment is shoveled into an array of (say) four 20-liter buckets until they are just filled to the brim. Each shovel-full goes into a different bucket than the previous one (like dealing a hand of cards) to help ensure that the samples are well mixed. As sediment is transferred into the buckets, all clasts with median diameters larger than 64 mm (this limit could be altered to satisfy the analyst) are separated and placed on a plastic sheet. The total volume of solids collected is thus the combined volume of the large clasts on the sheet, plus the bulk volume of the sediment in the four buckets, minus the bulk porosity of the sediment in the buckets. The first filled bucket is used to determine bulk porosity by adding measured amounts of water to the bucket until the sample is saturated to the lip of the bucket. The porosity is therefore the volume of water added, divided by the volume of the bucket. If the sediment was originally wet, the field capacity of the sediment would be added to the measured porosity, but this value is usually negligible «2% by volume) for coarse sediment. The volume of sediment in the bucket is simply the bucket volume minus the added water volume, and the value is assumed to apply to all the other buckets as well. The contents of the first bucket are then discarded. The contents of the second bucket are wet-sieved in the field to characterize the grain-size distribution of clasts smaller than 64 mm in diameter. The volume of solids caught on each sieve is determined by volumetric displacement of water. The (now damp) sediment of each size class is added to a known volume of water in a graduated cylinder; additional measured volumes of water may need to be added to ensure that the sediment is fully submerged. Both the volume of water and the total volume (water + sediment) are recorded. The field capacity also affects these measurements, and Booth et al. (personal communication) determined the correc tions for field capacity by re-creating field conditions in the laboratory and meas uring the amount of retained water: (continued) Chapter 4: Channel transport and storage 77

continued: Box 17 Sieve size Sieve size Sieve size (mm) Multiplier (mm) Multiplier (mm) Multiplier 0.031 0.40 0.50 0.62 8 0.87 0.062 0.43 1.00 0.67 16 0.90 0.125 0.52 2 0.72 32 0.92 0.25 0.55 4 0.82 64 1.00 The volume of solids is then calculated as the total volume less the volume of water, multiplied by the tabulated correction for field capacity. Where the bed material is coarse enough to require the field sieving method, grains of less than 0.5 mm in diameter usually need not be sieved. All clasts that had been set aside because they were larger than 64 mm are measured by one of two methods. Either the three diameters of each clast are meZl.sured and the volume subsequently calculated as V = 4/3 7t «DJ+D2+D3)/3)3; or the amount of water displaced by the clasts is measured in a calibrated bucket. The total volume of the sample is then the sum of the individual large-clast volumes and the sediment volume in the four buckets. The proportion of each grain size relative to this total volume can now be calculated. The contents of all other buckets are discarded, having been collected only to increase the total sample volume and thus improve the sanlpling of clasts larger than 64 mm. The grain-size distribution within each of the buckets is assumed to be the same because they were all filled concurrently as the sample was collected, rather than sequentially. With this technique, three people can collect and analyze a sample of about 80 to 100 liters in two hours. To minimize sampling errors, the largest single clast in a sample should occupy less than 1% of the total sample (Church et al. 1987); a more relaxed criterion of 5% of the total sample (Mosley and Tindale 1985) is sometimes unavoidable. In typical gravel-bedded rivers, an 80-liter volume usually achieves the first criterion for clasts up to about 100 mm diameter and the second criterion for clasts up to about 180 mm. For greater precision or larger sediment, earth moving equipment is almost certainly required. For most purposes the added time and expense are not justifiable. Modifications to the method might include drying the sediment on tarps to allow dry sieving. and weighing instead of volumetric measurements, aggregation of random samples over an area of uniform surface characteristics to provide a better estimate of the average distribution, and field-sieving only to 11 mm and sub sampling 5 to 10 kg of the finer sediment for later sieving in the lab (T. Lisle, USDA Forest Service Pacific Southwest Research Station, personal communica tion). 78 Chapter 4: Channel transport and storage example, randomized point counting (Wolman 1954, Kellerhals and Bray 1971) will produce an adequate grain-size distribution if the surface layer is the main concern, and for cobble-sized sediment this is the only feasible method. Kellerhals and Bray (1971) demonstrate that areal point counts provide values equivalent to weight proportions measured by sieving. Mosley and Tindale (1985) explore bed variability in braided gravel streams and report that about 1700 points need to be sampled to provide grain-size characterizations accurate to within 20%. This example represents an extreme case because the distribution of grain-size clusters in the sampled reaches was particularly complex. Results from several pebble counts over the same bar can be used to estimate the sampling variance for a particular site, and this value can then be used to calculate the number of samples required to achieve a desired level of precision. In the case of sand or gravel, grab samples may be taken from the surface or substrate and sieved. Where gravel is being sampled, a common rule of thumb is to collect large enough samples that the largest stone represents less than 5% of the total sample weight. Church et al. (1987) provide guidelines for selecting sample sizes for several types of measurements. Again, many samples are usually required if characterization is to be precise: Mosley and Tindale (1985) report that 50 30-kg samples were needed to define mean grain size to within 20% at their extremely complex gravel-bedded river site, and that refinement to within 10% required 228 samples. Ashmore et al. (1989) required about 10 grab samples from a sand-bedded stream to achieve a ± 10% precision. Fortunately, the awesome labor requirements suggested by these studies are not necessary for every river. Not all rivers are as morphologically and texturally complex as those on which the methodological issues have been explored. There is also room for reducing the sampling problem by areal stratification, hydraulic reasoning, and clear definition of why textural information is required. For example, if one is concerned with bedload transport rates, there is little use in augmenting the variance of gravely bed material by sampling thin layers of suspendible sand that were stranded during the previous flood recession. If the concern is the composition of spawning gravels, there is little point in trying to describe the textural variability of channels and bars inundated only at high flows. One also needs to question the level of precision necessary for the particular purpose. Given the inherent uncer tainty· of bedload transport equations, for example, there is little reason to exhaus tively sieve bed-material samples to establish a precise Dso to use in an equation. For some sediment routing problems, judicious selection of reaches for sampling can also reduce the problem of spatial variability. Where possible, one might select the straighter, simpler reaches upstream and downstream of a complicated braided reach to characterize bed material and make transport calculations. The straighter reaches will also be the only channels for which one can reconstruct reliable hy draulic information for transport calculations. Of course, it will not always be Chapter 4: Channel transport and storage 79

Box 18 Determining the required sample size for a pebble count 1. Measure the intermediate axes on five sets of 20 clasts, and categorize each into class intervals with a logarithmic spacing proportional to .y2 mm, as shown below. To simplify calculations, we will use the logarithmic scale known as the 4> scale, which is used by geologists to describe grain size. A 4> value is calculated as 4> = 1.44 In(mm) .

minterval mm interval minterval mm interval minterval mm interval >0 <1.0 mm -3.0 -3 .5 8.0-11 -6.0 -6.5 64-90 0.0 to -0.5 1.0-1.4 -3.5 -4.0 11-16 -6.5 -7.0 90-128 -0.5 -1.0 1.4-2.0 -4.0 -4.5 16-22 -7.0 -7.5 128-180 -1.0 -1.5 2.0-2.8 -4.5 -5.0 22-32 -7.5 -8.0 180-256 -1.5 -2.0 2.8-4.0 -5.0 -5.5 32-45 -8.0 -8.5 256-360 -2.0 -2.5 4.0-5.6 -5.5 -6.0 45-64 -8.5 -9.0 360-512 -2.5 -3.0 5.6-8.0

2. Identify the median axis class in each of the five sample sets

3. Calculate the mean (11) and standard deviation (0) of the medians (Mj), where n is the number of sample sets:

4. Decide to what accuracy you need to know the median grain diameter (Dso), and calculate the acceptable range as a percentage of 11 (% error). 5. The number of pebbles you will need to measure (N) to have a 95% confidence level of defining the Dso to within ±(% error) is approximately: 2 N _ ( 1.960 / 11 ) ~ % error 2 (method by R. Thomas and 1. Baldwin, USFS Pacific Southwest Research Station, personal communication) possible to reduce the problems caused by textural complexity in these ways, but there is little to be gained from rejoicing in the statistical difficulties that may arise from the complicated nature of some channel reaches. Much of the variation reported by Ashmore et al. (1989) and Mosley and Tindale (1985) results from predictable variations in bed material in different parts of a 80 Chapter 4: Channel transport and storage

Box 19 Example of a calculation of required sample size for Dso Five sets of 20 samples are drawn randomly from a population with an unknown median of -3.5 ~ (11 mm) and an unknown standard deviation of2 . 0~.

Size class Size class (~scale) (mm) set 1 set 2 set 3 set 4 set 5 pooled >0 < \.O 2 I 3 0.0 to -0.5 \.0-1.4 2 -0.5 -1.0 1.4-2.0 2 3 -\.0-\.5 2.0-2.8 2 2 4 -1 .5 -2.0 2.8-4.0 4 2 2 2 II -2.0 -2 .5 4.0-5.7 3 2 3 2 2 12 -2.5 -3.0 5.7-8.0 2 4 4 12 -3.0 -3.5 8-11 4 2 4 5 16 -3.5 -4.0 11-16 1 4 7 -4.0 -4.5 16-23 2 3 -4.5 -5.0 23-32 2 5 2 10 -5.0 -5.5 32-45 4 2 7 -5.5 -6.0 45-64 I -6.0 -6.5 64-90 2 3 -6.5-7.0 90-128 3 4 -7 .0 -7.5 128-181 2

median~: -3.25 -4.75 -2.75 -2.75 -3.25 -3.25 mean of medians: -3.35, standard deviation: 0.822

Desired accuracy: 95% confidence of being within 0.25 ~ of the true median. As suming the median is near -3.35, this would require an estimate accurate to ±7.5%. Substituting these values into the equation: N = {1.96(0.822 3.35)}2 ~ = 65 0.075 2 Thus the median calculated by pooling the five subsets into a sample of 100 will provide an adequate estimate of the population median, and produces an estimated median of -3.25 ~. To test the adequacy of the estimated minimum sample size, we selected ten sets of 65 samples. These produced the following estimated medians: -3.25 -2.75 -3.75 -3.75 -3.25 -3.25 -3.75 -3.75 -3.75 -3.25

As expected, nine of the ten samples are within 0.25 ~ ofthe population median. channel, and variance can be greatly reduced by taking these patterns into account while sampling. There is a characteristic spatial pattern of grain sizes in meandering Chapter 4: Channel transport and storage 81 streams with point bars (figure 20). Particle size is usually greatest in some deep pools and near the head of a point bar, and it declines towards the tail of the bar and away from the convex bank. The distribution of grain sizes is more complex on mid-channel bars in braided reaches and on ephemeral patches of sediment, but there is usually some recognizable pattern, such as coarsening on the upstream nose of the bar and fining downstream along contours; these can usually be defined easily in the field. The patterns are usually stronger in the pavement than in the substrate. Recognition of such patterns allows the investigator to choose compara ble sites among reaches for sample collection without having to map the channel. Smith (1974) describes some trends and complexities to look for in bar sedimentol ogy. If an average size of material for a whole channel reach is needed, stratified sampling combined with weighted averaging of sieving results is necessary. If one is simply defining spatial or temporal trends, sampling of equivalent sites is ade quate.

bed flow direction at mid · point of line

contour interval on channel bed = 20cm; low points shaded

particle size (d 75) in ITlT1

channel edgu undercut unless dashed

Stipple denotes 1982 point bill revogetated by 1984

o 50m I

Figure 20: Contour map of bed elevations in a meander on Waireka Stream, New Zealand. The contour interval is 20 cm. Pools are shaded black, and the stippled zone is the point bar revegetated within 2 years after aflood. Arrows denote near-bedflow directions, and numerals are D-s values in mm. Source: Carson and Griffiths (1987). 82 Chapter 4: Channel transport and storage

Sample collection is particularly difficult in inundated parts of the channel. A widely used device for such sampling is the McNeil sampler (McNeil and Ahnell 1960). The original design is a steel tube 16 cm in diameter that is encircled by a trough. The tube is driven into the bed, and sediment from inside the tube is scooped into the collar trough and removed for sieving. Common variations in the design include increased tube diameters to sample larger pebbles, and elongated tubes for use in deeper flows. Although some fine sediment is lost during scooping and the total sample size is small, Young et al. (1991) find that of commonly used methods, the McNeil sampler provides the most consistent and accurate estimates of bed composition, provided that the maximum grain size is less than about one-third of the sampler diameter. The McNeil sampler cannot provide information on layering of grain sizes within a sample. If the stratigraphy of the bar is to be evaluated, freeze-core samples (Walkotten 1976) may be collected, although the complexity of this process fre quently limits the number of samples collected. Because the volume that can be collected by freeze-coring is small, the technique is best used for fine-grained sediments. Large pebbles frozen to the core at one end protrude beyond the mean dianleter of the frozen column, and Young et al. (1991) find that freeze-core sam plers tend to over-estimate the proportion of large grain sizes. The problem of adequately accounting for spatial patterns of grain-size variation must also be dealt with when using McNeil and freeze-core samplers, but the variance in subsurface texture is likely to be less than that ofthe pavement layer. However, we know of no thorough analysis of the sampling requirements for these methods. Identifying temporal trends in grain-size distributions presents additional prob lems. It is possible to relocate sampling localities quite precisely. However, tempo ral variation caused by storm sequences and seasonal changes is often so large that subtle trends due to land-use changes are difficult to detect using repeated samples. In this case, it is even more crucial that spatial variations be documented so that an adequate sampling strategy for long-term monitoring can be designed. Comparing grain-size distributions of sediment from analogous locations along the channel can reveal discontinuities in sediment transport through a drainage basin. Absence of coarse grain sizes at downstream sites implies either that they are being reduced in size by abrasion or breakage, or that the stream is not able to transport those grain sizes beyond a certain reach. The importance of abrasion and breakage can be evaluated by tumbling-mill tests (Collins and Dunne 1989), or sometimes by comparison of particle liiliologies between sites (Dietrich and Dunne ·1978). Where coarser particles accumulate due to lack of transport competence, channel and valley morphologies usually show evidence of chronic aggradation: channels are often braided or are wide and shallow; channels show evidence of rapid shifting; bars are well developed; some riparian plants and structures are partially buried; depositional fans may be forming on the valley floor; there is an Chapter 4: Channel transport and storage 83

abrupt break in slope between the valley floor and adjacent hillslopes; and overbank flooding is relatively common.

4.4 Initiation of bed-material transport Analyses of the potential effects of altered flow regimes on sediment transport and channel morphology usually require evaluation of the mobility of channel-bed particles, and mobility calculations are als,? useful for predicting the likely effect of sediment inputs to a channel. Usually such calculations focus on determining what grain sizes can be moved by what discharges. Close inspection of a gravel- or cobble-bedded stream reveals particles ranging in size from sand to cobbles that are resting against and on top of one another. For any of these particles to be moved, the short-term combination of lift and drag forces in the near-bed flow must be sufficient to rotate the particle out of its pocket between other grains and over its downstream neighbor. The magnitude of this short-term combination of forces is thought to be at least roughly related to the mean flow condition near the bed, and is affected by the particle's shape, imbrication, exposure, and size relative to that of its neighbor. The geometry and dynamics of the initiation of particle transport have been described by Komar and Li (1986) and Wiberg and Smith (1987) and are reviewed by Richards (1990). The traditional method for predicting the initial motion of a bed particle (Shields 1936, described by Graf 1971, pp. 94-97) analyzes the effect of the time-averaged shear stress in the near-bed flow on the drag and lift forces that rotate a particle out . of a pocket formed by neighboring grains of the same size. This method, which is applicable to sand beds and to other surfaces on which particles are of nearly identical sizes, yields the equation:

'te = pgheS = 0.06(p - p,)gD (4.5) where p and p, are the densities of fluid and sediment, respectively, g is the gravita tional acceleration, S is the water-surface gradient, he is the flow depth at initiation of motion, and D is the diameter of both the particle of interest and its neighbors. The value of'te in Equation 4.5 is the critical shear stress at which transport begins, and 'tc is linearly proportional to particle diameter. A dimensionless critical shear

stress, 'tc', can be obtained by multiplying 'te by a nominal area over which the stress is assumed to act and dividing the resulting force by the submerged weight of the particle. The result predicts 'te' to be constant for rough turbulent flow: S 't; = pghe = 0.06 (4.6) (p - p,.)gD

The equation was developed using measurements from laboratory flumes with uniform-sized sediment. 84 Chapter 4: Channel transport and storage

For many years the Shields equation, and modifications of it with different values of the empirical constant, were applied to gravel-bedded rivers as the basis for estimating the flow depth required to transport a particle or for determining whether a particle of a given diameter would be mobile in reaches with calculated mean depth-slope combinations. However, it eventually became apparent that the method gives umeliable results for many gravel-bedded rivers b~cause of the effects of grain-size sorting on the pocket geometry referred to above, and because of the larger-scale effects of topographic and textural heterogeneity on the spatial variabil ity of shear stress and its effectiveness in moving grains (Church 1978) . The spatial heterogeneities remain a problem, but the effect of pocket geometry has been analyzed recently by theoretical and field studies. The work confirms that the critical shear stress required to transport a particle depends not only on the absolute size of the grain but also on its size relative to surrounding particles. Particles coarser than the local average move at t-values lower than predicted by Shields, and particles finer than average move only at higher t-values than pre dicted. No consensus has yet been reached on the magnitude of the effect of sur rounding particles in reducing or even canceling the effect of absolute grain size. One claim, originating from field data (Parker et al. 1982) and theoretical analy sis (Wiberg and Smith 1987), is that all particle sizes on a bed are equally mobile in the sense that they all move at the same shear stress. Thus, following Parker et al. (1982), Equation 4.6 can be replaced by:

,; = pghcS = 0.09(~J-I.OO (4.7) (p - pJgD D50

for 0.045 < D1D50 < 4.2. Here D is the diameter of the particle of interest and D50 is the median diameter of the sub-pavement. For a fixed value of D50' changing D in Equation 4.7 affects both sides of the equation equally and the result remains (4.8) However, this relation was derived using transport measurements made after an appreciable portion of the bed surface was in motion, and the 'c solved for is that at which a small but finite amount of transport is taking place. Conditions at the onset of motion were not explicitly measured, so the equation is best used to describe transport conditions after motion begins. Field measurements by Andrews (1983) also address only those transport stages for which an appreciable portion of the bed is in motion. More recent studies have suggested that although the dependence of the critical shear stress on particle size is reduced below that predicted by the Shields equation for heterogeneous beds, it is not adequately described by equations that predict equal mobility of different particle sizes. There is not yet a consensus about the degree of reduction, which should be reflected in the exponent of Equation 4.7. A Chapter 4: Channel transport and storage 85

Table 7 - Equations/or initiation a/motion. 'c (dynes-cm- 2) is calculated/or particles a/size D; (cm) lying in a pavement or on a substrate with median grain size D50 (cm) or geometric mean diameter D" (cm).

Source Equation 0 50 type Method'"

Shields 1936 "t c = 0.056(p, - p) gD;

Miller et al. 1977 "t c = 0.045(p, - p) gD;

Parker et al. 1982 "t c = 0.09(p, - P)gD50 subsurface bedload

Oiplas 1987 "t = 0087 (p - p) gDO.0 6 DO. 94 subsurface bedload c' S I 50

Parker 1990' "t ="t (p - p) gDO. 10 DO.90 surface bedload c rg S I g

Komar 1987a" "t = 0045 (p - p) gD°.35 DO. 65 subsurface clast cos I 50

Andrews 1983 "t = 0083 (p - p) gDO\3 DO.87 subsurface clast c ' S I 50

Ashworth & Ferguson "t c = 0.072 (p, - p) gD;0.35 D~t surface bedload 1989

"t = 0054 (p - p) gD033 DO.67 surface bedload c' , I 50 "t = 0087 (p - p) gD008 D9·92 surface bedload c' S I ,0

Komar & Carling 1991 "t = 0059 (p - p) gD°.36 D064 subsurface clast c' S I 50

"t =0039(p _p)gD°.l8Do.82 subsurface clast ('. S I 50 Costa 1983 deposit "t c = 26.6 D/ 21

Williams 1983 "t c = 12.9 D;1.34 deposit

* D. is the surface geometric mean particle diameter and is calculated as: In D. = ~ Fi In Db where F; is the fraction of particles in size class i and having a geometric mean of D;

"t rg = 0.836 (D!( / D50mh ) -0.905 is the reference dimensionless shear stress for Dg ** Equation was derived using surface grain parameter but was applied using subsurface. *** Method: bedload Initiation defined by the occurrence of a small but finite amount of transport. clast Initiation threshold defined from the largest particle moving at given ,. deposit Initiation threshold defined from the largest particles deposited during an event with given, at crest stage. value of 0.0 would convert this equation to the form of the original Shields relation, while a value of -1.0 would indicate that, for a given 0 5°' all grains on the bed would move at the same value of "to Many researchers have developed similar equations for particular channels (table 7), and empirically defined exponents vary between about -0.6 and -1.0. Most workers use the median grain diameter of the 86 Chapter 4: Channel transport and storage subsurface sediment as the reference grain-size parameter, but there is little physical' support for this unless the pavement is thin and discontinuous. Instead, the particle size of most direct relevance is that immediately downstream of the grain in ques tion, and this is better described by a measure of the surface grain-size distribution. Because of this dependence, Parker (1990) has altered his equation to use a surface grain parameter rather than the subsurface median. Differences between equations for initiation of motion result for several reasons. First, the equations of Costa (1983) and Williams (1983) are based on the cessation of motion, which usually occurs at a lower shear stress than initiation of motion. These equations take account of neither the conditions of entrainment nor the range of particle sizes in the depositional zone, and they are best used only as indicators for conditions of deposition. Second, the other studies identify initiation of motion in different ways. For example, Parker et al. (1982) and Ashworth and Ferguson

(1989) use a 'tc calculated from field measurements for a particular finite transport rate, while Andrews (1983) and Komar (1987a) define 't c by determining the size of the largest particle in motion at a particular value of't. Other differences reflect contrasting definitions of't. Parker et al. (1982) use the average basal shear stress defined in Equation 4.8, while Ashworth and Ferguson (1989) calculate the local basal shear stress from near-bed velocity measurements. Where relations are parameterized by the subsurface median grain diameter, equa tions depend on the relation between this value and the surface grain diameters that more closely control initiation of transport, and coefficients are likely to vary as a function of the ratio between surface and subsurface medians. The field measurements of Figure 19 can be used to illustrate the differences in sediment transport behavior predicted by the various equations for initiation of motion (figure 21). In this case, the channel has a median grain size of 56 mm in the surface layer and 17 mm in the subsurface. Straightforward application of the Parker et al. (1982) equation suggests that all particles in the surface layer remain fixed until the shear stress exceeds 248 dynes/cm2, and then all particles of the pavement are intermittently mobile and equally likely to move. This situation is shown by the vertical step in Figure 21, which indicates the proportion of the bed material that is predicted to be mobile at any shear stress. By contrast, the Shields equation predicts that larger size classes become mobile at progressively higher shear stresses over a very wide range of shear stresses. The various equations developed by Ashworth and Ferguson (1989) define a range of intermediate condi tions, but with critical shear stresses several times larger than that given by the Parker et al. (1982) equation because Ashworth and Ferguson use the median grain size of the surface material rather than that of the substrate. The Parker (1990) equation is similarly dominated by the subsurface grain size because inclusion of't'll (see footnote of Table 7) reduces the influence of the surface grain-size parameter. The Komar (1987a) equation also uses a subsurface grain parameter and so gives Chapter 4: Channel transport and storage 87 roughly the same answer as the equations of Parker, although it suggests that various size classes move at a slightly wider range of shear stresses.

100r-----,,.~---.-----,~~----rr--~-.~~~--=---, I Parker 1990 / )-../ Shields (/) rna> 80 Parker et II / / / /: .L:...... _ - al.1~1· ...... 0 cO a>E I / // I E» 60 ~;:; / I / / I roC o.Jg c)E 40 / I /{AShworth & em Komar/ . Ferguson . a>e 1987 / V ~I <..).- /. - '- 20 I a> / / relatively /. relatively CL V . mobile bed stable bed OL~=}--/~~~'_/~--~~/--~~~~ o 200 400 600 800 1000 1200 Shear stress (dynes/cm2)

Figure 21: Percent of a channel-bed pavement (represented in Figure 19) that is intermittently mobile at various shear stresses, according to six of the equations listed in Table 7. The shapes of the curves reflect both the equation and the shape ofthe grain-size frequency distribution.

Thus, even if one accepts that the Shields equation should not be applied to gravel-bedded streams because it assumes that particles are of uniform size, the alternative equations produce conflicting results. Parker's approach suggests that all bed particles are mobile within a shear-stress range of ±20% ofthe 'c of the median size of the pavement sediment; other equations predict general mobility within

ranges of ±20-50% of this 'tc; and still others predict ranges about quite different values of 'c' Given these differences in predicted behavior, testing which equation is most useful would seem to be a straightforward task. However, few data sets exist, and most have been used to derive the various equations. Because none of the equations has been adequately validated for general applica tioll, none can be used confidently to predict particle size sorting or conditions required for initiation of bedload transport. The only general consensus is that in poorly sorted gravel beds, the various particle sizes move at a narrower range of shear stresses-and, for coarse particles, at lower values-than is indicated by the traditional approach of extrapolating the Shields equation (Equation 4.5) into conditions for which it was not developed. 88 Chapter 4: Channel transport and storage

The study by Ashworth and Ferguson (1989) suggests that size selection be comes less important as transport intensity increases. Because the transport intensity is low as the bed begins to move, this dependence implies that initial entrainment from the bed is likely to be size-selective. As the proportion of the bed in motion increases, however, the size distribution of the bedload increasingly approaches that of the bed (Figure 2 of Ashworth and Ferguson 1989). Conditions then more closely reflect those used to derive the Parker and Andrews equations, and these 'become more appropriate. This pattern is demonstrated by a comparison between the equations' predictions and measurements of the largest grains in transport as a function of basal shear stress for a small Oregon stream (figure 22). Parker et al. (1982) predict that the largest and smallest grains both will begin to move at about

.~ 1 Costa Ashworth & /1.. • 400 Ferguson / . Shields (mbbileJj / • • • ••• • Andrews • --- / NE / / •• ___ ------Komar & • --- Carling ;/ / . .::::::. .. -parker 1990 ~Q) 300 ~ -;- Parker et al. 1982 :; / / • --- en /. --- · .... i<.omar1987------. ~ I. ~., • ~ 200 I .. ( • ./'" --- m I ~ / c7j I 1. 1/.. " : I • 100 1/· : I· / 1/ O~/~~~--~~~~~~~--~~~~ o 2 4 6 8 10 12 14 Grain size (em)

Figure 22: Comparisons of shear stress for initial motion of particles of various sizes predicted using initial-motion equations and those measured by Milhous (1973) at Oak Creek, Oregon. Measured values represent the maximum particle size transported during f1ow~ of the indicated total shear stress. Source: equations superimposed on figure from Komar 1987b. Chapter 4: Channel transport and storage 89 the same shear stress, and their predicted threshold shear stress roughly corresponds to the stress at which the measurements show the surface Dso becomes mobile. In contrast, Shield's (1936) equation more accurately predicts the stress at which grains much smaller than the Dso begin to move, as do Komar's equations (1987a and Komar and Carling 1991), which were derived using information on initation of motion. The data set shown in Figure 22 was used in the derivation of the Parker, Diplas, and Komar equations. It is useful at this point to review whatis meant by the term "equal mobility", which is now widely used in the literature on sediment transport in gravel-bedded streams. The term is employed in at least two ways. First, it is used to describe a condition in which each grain on a channel bed has an equal probability of being mobilized at a given shear stress. The curve predicted by the Parker et al. (1982) equation in Figure 21 represents the extreme of this condition, where all particles become mobile at a particular threshold value for shear stress. This type of equal mobility is likely to be approached most closely where much of the bed is in motion, because particle impacts and undermining are likely to enhance the mobil ity of coarse particles beyond the effects predicted by fluid stresses. Others use "equal mobility" to refer to a condition in which the transport rate of bed material size fractions is proportional to their representation in the bed, so that the grain-size distributions of the bed-material load and the bed material in a reach are the same. This condition requires that each grain size travel at the same average speed through the reach. This average speed depends on both the frequency of motion and the distance a particle travels after each excitation. In other words, if grains of each size have an equal probability of entrainment during the most signifi cant transport events, then grains of each size would also have to travel the same average distance after each entrainment if the size distribution ofthe bed material is to equal that of the bedload. Conversely, if small grains are more readily entrained than large grains, they would have to move, on average, a shorter distance per excitation. If, after a significant transport event, all grain sizes have travelled about the same distance, then any size dependence in entrainment probability has been compensated for by a size dependence in travel distance per entrainment, and equal mobility can be assumed. Unfortunately, the available data for distance of transport are at least as variable as those for entrainment. Laronne and Carson (1976) painted rocks in the size range 4-256 mm and injected them into a streambed during snowmelt. At the end of the melt season, approximately 5% of the marked particles were recovered. Although their distance of transport is crudely negatively correlated with particle weight, there is a ten-fold range in distance traveled for a given grain size, and the location of deposition also depends on channel shape, bed topography,. and sedimentary structure of the bed. Ashmore and Ferguson (1989) also find weak negative correla tions between particle weight and distance of transport, but these patterns also are not strong enough to define a general relation. 90 Chapter 4: Channel transport and storage

Hassan et al. (1991) placed magnetically tagged particles (30-180 mm) on the bed of small gravel-bed streams in Israel and traced their movement during single floods. There is a rough correlation between the mean distance traveled (10-150 m) and the stream power per unit area for the peak flood discharge, but storage of particles in bars complicates the relationship as the travel distance im;reases further. No simple relation is evident between particle size and distance moved over these short reaches. Hassan et al. (1992) combined travel distances measured on particles from sand to cobbles in a wide range of channel types and found the distances (1- 250 m) to be weakly correlated with stream power per unit area in excess of that required for mobilization of the bed, which was assumed to be a function of the median grain size of the subsurface material. Gravel particles moving over a sandy bed traveled much further than those on a rough bed for the same excess stream power. On the gravely beds, however, grain size again has only a weak influence on travel distance, at least as measured over short time periods and distances. So, even if all grain sizes do not travel at the same rate over long distances and time periods, the sorting appears to be inefficient over short distances, and the size distribution of the bedload can be assumed to be approximately equal to that of the bed. Given the complications discussed above, it is wise to make some local field observations as a check on the predicted critical shear stress. Maximum grain sizes that are mobile during specific floods can be estimated by measuring the largest particles that were obviously rearranged or imbricated on gravel bars, or that were deposited over new organic debris or trash. However, one must be careful to deter mine whether evidence for mobility of a large particle reflects the shear stress in uniform flow in a straight reach or whether it is influenced by shear stress due to local flow accelerations around steep bars, bedrock obstructions, or organic debris. We have also used painted rocks and scour chains to identify the critical discharge at which gravel pavements were scoured. These measurement techniques can be incorporated into a reconnaissance-level sediment budgeting effort if fieldwork is scheduled for before and after the flood season. In other cases, the record books of equipment operators have proved useful for identifying mobilizing discharges.

4.5 Determination of scour depths Evaluation of the quantity of sediment available for transport requires some estimate of the depth to which scour is possible in alluvial channel beds. Unfortu nately, this problem does not yet have an adequate solution. The most successful predictions of scour have dealt with the condition of channel beds after sediment input has been reduced by dam construction. In these cases, sediment-transporting flows are capable of carrying off the finer components of the bed and thus progres sively coarsening the bed until the remaining particles are too large for further winnowing. The depth of scour from winnowing can be estimated by determining Chapter 4: Channel transport and storage 91 the proportion of non-transportable sizes in the bed material and calculating the volume of finer sediment that must be removed to leave a bed surface composed solely of the larger sizes. This procedure is described by Borah (1989). The more general case is where sediment transport from upstream replaces sediment mobilized at a site. During floods, the bathymetry of a channel may change through rearrangement of the channel bed, or because the bed to some finite depth is in transport, or because sediment transport at a site is greater or lesser than the instantaneous contribution of sediment to the site. A cross section measured during a flood peak thus may be quite different than that measured during rising and falling limbs, and Leopold et al. (1964) cite measurements demonstrating that one quarter of the flow depth during a flood peak on the San Juan River was accommodated by scouring of the bed. The minimum magnitude of such changes can sometimes be estimated by comparing channel cross sections before and after storms, but this will not reveal the magnitude of the cycle of changes that may occur during a single flood. Current-meter records "from gauging stations include repeated measurements of flow depths and gauge height (and therefore of bed-elevation cross sections) during and between floods. Superimposition of many cross sections at a gauge can indicate the range of scour and fill and the conditions under which each occurs. Passage of bedforms along a channel bed may also produce changes in at-a site depth that may approach several meters in the case of sand dunes in large rivers. The magnitude of this effect can be estimated simply by measuring the height of migrating bedforms. Observations during storms show that parts of a channel may experience scour while other reaches are aggrading. Andrews (1979), for example, found that reaches that were narrower and deeper than average along the East Fork River in Wyoming tended to scour, while wider and shallower reaches aggraded. Because of the importance oflocal conditions in determining scour intensity, and because the propensity for sediment transport at a site depends strongly on the characteristics of the sediment present, there is little prospect for development of a general equation for predicting scouring depths. However, relationships between scour and discharge have been defined for several channels, and we list them here to show how local data might be analyzed, rather than as tools for predictions at other sites. Leopold et al. (1966) measured scour depths during peak flows in an ephem eral channel in New Mexicp and found that mean scour depth across a channel (d" meters) varies with discharge per unit channel width (q, m3s· l m· I ): d, = O.lOq05 (4.9) Because the channel is sand-bedded, a large portion of this effect is expected to be caused by bedform migration: the channel will appear to deepen considerably as the trough of a sand-wave passes by. Carling (1987) compiled mean cross-sectional scour measurements from several gravel-bedded channels to define the relations: 92 Chapter 4: Channel transport and storage

d, = 0.043 q 027 r= 0.19 (4.10)

r= 0.54 (4.11 )

where dj (meters) is the depth of fill. Maximum values in both cases (dlllaT, meters) were described by the relation: r= 0.19 (4.12)

Hassan (1990) compares scour and fill depths predicted by Carling's equations to measured values from a gravel-bedded channel in Israel and finds that predictions are generally good to within a factor of 3 at that site. Equations may thus be useful for suggesting the order of magnitude of scour and fill, but calculated values should not be used without validation by some field measurements. Clearly, scour or fill will not occur until the bed surface is mobilized, so even locally derived equations should be applied only to those events for which bedload transport occurs.

4.6 Sediment transport rates in channels Natural or land-use-related environmental changes commonly affect channel morphology and sedimentation by altering sediment transport rates, so it is often necessary to compare transport rates in a reach before and after a proposed change in land use or flow regime. Transport calculations are also useful for identifying the locations of potential aggradation or degradation. Formulas exist for predicting sediment transport capacity, but their value is best summarized by the compilers of a manual on sedimentation engineering (Vanoni 1975, p. 190): "Unfortunately, available methods or relations for computing sediment discharge are far from completely satisfactory with the result that plans for works involving sediment movement by water cannot be based strongly on such relations. At best these relations serve as guides to planning and usually the engineer is forced to rely strongly on experi ence and judgment in such work. " The compilers might also have mentioned the value of having a few field measure ments 'as an independent test of calculated results. Given the emphasis of the manual, the compilers' pessimism seems to refer mainly to sand-bedded alluvial channels, which have been the focus of most sediment transport research. The problem is even less tractable in steep and rough mountain streams with gravel and cobble beds, where the hydraulics are exceedingly complex and poorly defined, and where the mobile bed sediment may be discontinuous and coarser than the sediment used in constructing the largely empirical equations. All sediment transport formulas are based on the assumptions that discharge remains constant for Chapter 4: Channel transport and storage 93 long periods, that downstream variations of depth and velocity are minor, and that there is an infinite supply of sediment of grain sizes represented by some compo nent of the bed material. These conditions obviously are not well-met in many mountain channels. Although modifications of these restrictions are being intro duced, the work remains in a research phase and usually requires measurements and computations that are too complex for routine applications on small-scale sedimen tation problems. Therefore, one must rely on empirical field methods for under standing sediment transport and for checking the results of transport equations before using them for semi-quantitative explanations of present conditions and guides to estimating future changes.

4.6.1 Sediment transport equations Sediment transport equations generally compute only the bed-material load because the washload component depends strongly on sediment supply; washload must be evaluated separately by sampling, construction of sediment rating curves, and calculation of annual yields on the basis of flow-duration curves. Those equa tions developed for gravel-bedded channels are concerned primarily with bedload transport because this tends to be the transport mode for most of their bed material (table 8a). In contrast, equations developed using data for sand-bedded channels often calculate both the bedload and the bed-material suspended load (Table 8b), because sand-sized sediment is commonly partitioned between these modes. The latter are referred to as "total-load equations" even though they do not consider washload. Some total-load equations require separate analyses for bedload and bed material suspended load, and these can be used for both bedload and total-load calculations. Graf (1971) and Vanoni (1975) review formulas for bedload and bed-material load. Most of these equations are based on calculations of excess tractive force (the difference between the fluid shear stress and the critical shear stress required to initiate motion), velocity, or discharge, and they range from the totally empirical to those heavily based in theory. However, all employ some measure of empiricism in the definition of basic relationships, and because of this, each equation is best suited for conditions similar to those assumed in its development (table 8). Thus, no single equation has emerged as the definitive transport formula. Which formula is appro priate for a particular application depends on the type of bed material present and the channel characteristics. Data requirements vary slightly between formulas, but most require information on channel gradient, depth, width, and sediment character. A few total-load equations, such as the Modified Einstein and Toffaleti methods, require measurements of suspended sediment. Incorporation of the critical shear stress fo{ initial motion into the transport formulas introduces the uncertainties reviewed in the earlier discussion of initiation of transport. 94 Chapter 4: Channel transport and storage

Table 8 - Sediment transport equations. Where possible, references are given that discuss or illustrate the use ofthe equations. Where these are not available, only the primary source is given. A. BEDLOAD Sediment Grain-size Data*

Equation Code D50(mm) mixture fl ch Reference Bagnold 1956 B56 Bagnold 1956 Bagnold 1980 B80 1.1 ? uniform x Bagnold 1980* * Diplas Di 54 mixture x Diplas 1987 du Boys/Straub Du 0.1-4? uniforin? x Vanoni 1975, Brown 1950 Einstein, bedload Eb 0.3-29 uniform? x Vanoni 1975, Einstein 1950 Einstein/Brown EB 0.3-29 uniform? x Vanoni 1975**, Brown 1950 Engelund/Fredsoe EF Nakato 1990** Kalinske Ka Graf 1971, Kalinske 1947 Meyer-Peter MP 3-29 uniform x Vanoni 1975 Meyer-Peter/MUlier MM 0.4-29 mixture x Vanoni 1975 Parker 1982 P82 54 mixture x Gomez and Church 1989, Parker et al. ·1982 Rijn, bedload Rb 0.2-2 mixture? x Rijn 1984a Rottner Ro Rottner 1959 Schoklitsch 1934 S3 0.3-5 mixture? x Vanoni 1975, Graf 1971 Schoklitsch 1943 S4 sand? mixture x x Gomez and Church 1989, Graf 1971 Shields Sh 1.6-2.5 uniform x Vanoni 1975, Graf 1971 Toffaleti To sand mixture x x Vanoni 1975, Toffaleti 1969 Yalin 1963 Ya 0.3-29 mixture x Gomez and Church 1989 Yang 1984 Y84 2-7? uniform x Yang 1984

* primary source of data used. to develop equation: fl = flume, ch = channel ** reference corrects earlier version of equation

4.6.2 Comparison of transport predictions and measurements F ommlas to be used for a particular application should be selected on the basis of how well they have predicted loads in similar channels. Unfortunately, few equa tions have been widely tested in natural channels. Results of published comparisons are described in Table 9 and are summarized in Table 11 according to the size ofthe channel and the bed material. Channel characteristics are described in Table 12, and Table 13 identifies the sources of comparisons. Particularly useful are comp~risons by Vanoni (1975) for sand-bedded channels, Gomez and Church (1989) for gravel bedded channels, and White et al. (1978) for a range of bed material sizes. In many cases the test data were also used to construct the equations. Although this invali dates a general test of a formula's performance, it does not matter for our purposes: if the equation was designed to perform well on a particular channel, then it is likely that it will be appropriateJor similar channels. . Chapter 4: Channel transport and storage 95

Table 8 (continued) B. TOTAL BED-MATERIAL LOAD OR SUSPENDED BED-MATERIAL LOAD Sediment Grain-size Data* Equation Code DsoCmm) mixture fl ch Reference Ackers/White AW 0.04-5 uniform x Gomez and Church 1989, Ackers and White 1973 Ackers/WhitelDay AWD 0.4-14 mixture x Gomez and Church 1989 Ackers/Whitel AWS mixture x Gomez and Church 1989 Sutherland Bagnold 1966 B66 Graf 1971 Bishop BSR Graf 1971 , Bishop et al. 1965 Blench B1 sand Vanoni 1975, Blench 1966 Colby 1964 C64 sand mixture x Vanoni 1975, Colby 1964 Einstein, total Et 0.8-29 uniform x Graf 1971 , Einstein 1950 Einstein, modified Em Chow 1964, Colby & Hembree 1955 EngelundiFredsoe EF 0.9-\.8 uniform x Nakato 1990** EngelundiHansen EH 0.2-.9 Vanoni 1975 Graf Gr 0.1-3 mixture x x Graf 1971, Graf & Acaroglu 1968 Inglis/Lacey IL Vanoni 1975, Inglis 1968 Laursen La 0.01-4 mixture x Vanoni 1975, Laursen 1958 Maddock Ma 0.003-28 mixture x Maddock 1976 Rijn, suspended Rs 0.1-0.7 mixture x x Rijn 1984b Shen/H ung 1972 SH7 0.1-1.3 mixture x x Shen and Hung 1972 Shen/Hung 1983 SH8 0.08-2.3 mixture x Shen and Hung 1983 Toffaleti To sand mixture x x Vanoni 1975, Toffaleti 1969 Yang 1973 Y73 0.1-2 mixture x x Yang 1973 Yang 1979 Y79 sand mixture x Yang 1979

* primary source of data used to develop equation: fl = flume, ch = channel ** reference corrects earlier version of equation

The ideal equation for a particular application would be one that has been widely tested in similar channels; provides valid estimates for each channel of that type; provides consistent results for different operators on a single channel; and provides · valid estimates for all available data for a single channel. Many factors therefore must be considered to rank the effectiveness of predictive equations. Ranking is further complicated because published comparisons use different formats for reporting results. The most useful studies present the measured and calculated values either in tables or graphs, while others merely report the mean ratio between calculated and observed values. In addition, some comparisons reflect more than 100 individual samples, while others simply compare a multi-year Table 9: Tests of sediment transport equations in natural channels. The first digit of each three-digit number is an index of the number of comparisons for which the equation underestimates the measured load by more than a factor of two; the middle digit indicates the cases for which the values agree to within a factor of two; and the final value indicates the cases for which load is overestimated by more than two-fold. Channels are v:> described in Table 12; equations in Table 8; and sources in Table 13. 0-

A. GRA VEL-BEDDED CHANNELS

Small channels Medium channels 1. arge channels Braided channels Equation TW SL OA GO BG CP EF EL AA 10 ME VE SC SB TA CW SN KA WA OH RA Ackers/White 003 014 210 120 021 003 003 030 Ackers/White/Day 003 021 012 012 Ackers/White/Sutherland 003 003 012 003 Bagnold 1956 021 300 Bagnold 1980 141 051 020 010 510 120 120 100 100 100 Bishop et al. 003 003 Blench 003 Diplas 030 Duboys/Straub 003 003 120 003 Einstein, bedload 003 120 341 300 210 210 Einstein/Brown 003 030 003 003 030 Einstein, modified 021 9 Einstein, total 020 300 {j EngelundlFredsoe 300 300 ;;;-..., Engelund/Hansen 003 012 030 021 Graf 003 003 f:. Inglisn .acey 003 003 120 021 Laursen 120 300 9\::. ::! Maddock 100 ::! Meyer-Peter 030 003 300 012 ~ Meyer-PeterlMuller 300 002 300 215 300 300 300 300 201 030 010 030 100 300 411 ~ Parker 1982 030 002 003 150 120 ::! Rijn, bedload 210 021 ~ Rijn, suspended load 102 030 <::>..., Rottner 012 210 .... \::. Schoklitsch 1934 030 003 300 300 030 003 ::! Schoklitsch 1943 001 002 003 001 003 030 015 021 510 003 015 I::... c.., Shields 010 300 001 <:i Toffaleti, bedload 300 300 i:S Tooffaleti, suspended 030 003 ()Q Yalin 003 213 300 300 012 '" Yang 1973 120 012 010 Yang 1984 020 Q B. SAND-BEDDED CHANNELS (add 10 to each underlined digit: e.g.1 = 14) .g... Small channels Medium channels Large channels ~.... Equation WG MN· MU JC ID SK MY NI ML SJ RP FI RG CO MO RE PA AT MS MT ZA ~ Ackers/White 021 081 120 030 010 010 080 081 030 030 020 210 240 030 Bagnold 1956 020 003 001 120 020 300 030 Q \:l Bagnold 1966 030 030 030 100 100 030 030 011 120 231 300 ;:s Bagnold 1980 010 012 ;:s Bishop· et al. 012 012 010 060 150 120 210 200 300 210 300 ~ Blench 030 012 100 090 030 120 200 300 300 300 ~ Colby 1964 210 060 120 030 300 120 040 240 ;:s ~ DuboysiStraub 003 006 001 009 006 003 003 C:> Einstein, bedload 032 033 100 001 390 120 210 031 200 300 300 300 ...... Einstein/Brown 120 012 \:l 003 015 001 009 003 003 002 030 ;:s Einstein, modified 030 060 030 010 020 010 050 020 \:l.. Einstein,total 020 033 100 360 231 200 300 300 300 "" EngelundlHansen 020 072 021 030 OQl 061 030 030 010 020 120 060 030 C ~ Graf 003 012 001 003 030 020 210 120 300 OQ Inglis/Lacey 011 003 001 009 003 120 020 120 030 030 ~ Kalinske 021 006 001 220 450 020 200 110 300 Laursen 032 290 010 O£O 210 030 300 510 200 300 210 300 Maddock 330 060 060 150 330 Meyer-Peter 120 400 500 300 Meyer-Peter/Muller 141 850 030 020 750 630 111 300 040 300 300 300 Rijn, bedload 020 030 010 Rijn, total load 030 030 Rottner 003 051 060 030 002 012 021 030 Schoklitsch 1934 030 520 010 090 150 300 Shen/Hung 1972 030 120 060 060 030 300 300 510 300 Shen/Hung 1983 020 010 020 020 Shields 006 009 001 009 003 003 002 003 003 003 Toffaieti,bedload 111 001 Toffaleti,totalload 021 132 010 2JO 240 030 210 030 030 020 060 042 060 Yalin 003 003 001 003 003 002 300 012 210 Yang 1973 030 120 060 060 120 100 210 012 Yang 1979 030 030 030 021 \0 -..l 98 Chapter 4: Channel transport and storage

Box 20 How comparisons are noted in Table 9 The following method was developed for this application to allow data sets having different forms and precisions to be pooled to provide generalized descriptions of the accuracy of each equation: I. For each data set, determine which quantities are being compared, and if they are different (e.g. bedload is compared to total load), adjust the comparison by the likely difference as indicated by available measurements. 2. Create a histogram for each test where data are presented or for which a distribution can be inferred. Each histogram has only three categories: calculated value less than half the measured value, calculated value within a factor of two of the measured value, and calculated value more than twice the measured value. Each distribution is represented by up to three points, where: a. 1-3 samples: one point is assigned to the modal category b. 4-6 samples: two points are assigned to the modal category if it differs by more than one from the sub-modal category; otherwise, both the mode and sub mode receive one point. c. more than seven samples: three points are assigned, one to each "intact" quar tile, where an intact quartile contains between 25% and 50% of the observations. For example, a distribution of 3-4-1 would have one intact quartile in the first category, two in the second, and the fourth quartile is split between the first and third categories. If four quartiles are intact, the mode receives three points if it is the center category; otherwise it receives two points, and the adjacent cell 1. If only two quartiles are intact, the modal category receives a second point. 3. If only a mean value is reported, one point is assigned to the category con taining the mean; if only the mode is reported, two points are assigned to the category containing the mode. 4. Histogram values for different tests on the same river are summed and listed in Table 9. Each histogram is represt:nted by three numbers in the form "abc" (e.g. 081 for Ackers-White on Mountain Creek), where "a" (0 in this case) is the summed point value for underestimates, "b" (8 in this case) is the point value for estimates within a factor of two of the measured value, and "c" (l in this case) is the value for overestimates. Chapter 4: Channel transport and storage 99

Table 10: Histograms of modal values for sediment load equations. The sum of the digits in the histogram, divided by two, equals the number of different channels for which the equation has been tested. For example, an equation with the histogram 204 for medium gravel-bedded channels has been tested on three of these, has underestimated the result for one, overestimated it for two, and has not produced a mod,e within a factor of2 ofthe measured mode. Gravel-bedded channels: Sand-bedded channels: Equation small medium large braided small medium large Total and suspended load equations: Ackers and White 204 044 010 0120 0'60 280 Ackers, White and Day 022 004 Ackers, White and 004 004 Sutherland Bagnold 1966 020 480 251 Bishop et al. 004 024 260 800 Blench 002 222 060 800 Colby 1964 200 280 040 Einstein, modified 020 060 0100 Einstein, total 220 23 1 040 800 Engelund and Fredsoe 400 Engelund and Hansen 004 040 080 080 080 Graf 004 006 022 440 Inglis-Lacey 004 040 01 5 024 080 Laursen 220 060 440 10 0 0 Maddock 100 1 1 0 060 1 1 0 Rijn, suspended 022 Rijn, total 040 Shen and Hung 1972 040 060 800 Shen and Hung 1983 080 Toffaleti, suspended 022 Toffaleti, total 060 260 0120 Yang 1973 022 040 060 400 Yang 1979 020 060 002 Bedload equations: Bagnold 1956 220 024 020 240 Bagnold 1980 080 240 600 Diplas 020 duBoys-Straub 004 006 006 002 Einstein and Brown 022 004 020 006 006 044 Einstein, bedload 002 240 200 233 240 820 Kalinske 024 130 530 Meyer-Peter 022 202 220 400 Meyer-Peter and MUller 202 402 800 060 260 620 600 Parker 1982 022 022 620 Rijn, bedload 220 060 Rottner 202 022 040 044 Schokl itsch 1934 022 422 240 240 Schok1itsch 1943 008 044 204 Shields 220 002 006 006 008 Toffaleti, bedload 400 Yalin 204 202 006 004 404 Yang 1984 020 100 Chapter 4: Channel transport and storage

Table 11: Potentially useful sediment transport equations for channels of various types. Expla- nation ofsymbols and headings is shown below. Equation Table 10 Table 9 Table Independent type histogram histogram 9 ratio Rivers tests/total Rank Gravel-bedded, medium Bagnold 1980 b 080 1 122 0.20 4 2/6 B Gravel-bedded, large Engelund/Hansen 040 051 0.20 2 2/2 C Inglis/Lacey 040 14 I 0.33 2 2/2 0 Gravel-bedded, braided Meyer-Peter/Mu lIer b 060 070 0 3 3/3 A Sand-bedded, small Ackers/White 0120 0172 0.11 6 6/9 B Engelund/Hansen 080 o 143 0.18 4 6/6 B Laursen 060 2 132 0.24 3 3/5 B Shen/Hung 1972 040 150 0.20 2 2/2 C Yang 1973 040 I 50 0.20 2 1/2 E Toffaleti, total t 060 163 0.40 3 3/3 E Meyer-Peter/Muller b 260 9 14 I 0.42 4 10/10 E Schoklitsch 1934 b 240 560 0.45 3 5/5 E Sand-bedded, medium Engelund/Hansen 080 0222 0.08 4 717 A Shen/Hung 1972 060 o 150 0 3 3/5 A Einstein, modified 060 0120 0 3 3/4 A Ackers/White 060 o 19 I 0.05 3 517 A Maddock 060 I 170 0.06 3 3/6 A Blench 060 I 140 0.07 3 515 A Yang 1973 t 060 I 140 0.07 3 2/5 B Rottner b 040 090 0 2 3/3 C Einstein, total 040 591 0.40 2 515 0 Colby 1964 280 2130 0. 13 5 5/6 0 Toffaleti, total 260 6210 0.22 4 6/9 0 Bishop et al. t 260 4140 0.22 4 2/6 0 Kalinske b I 30 670 0.46 2 515 0 Bagnold 1966 t 480 2 120 0.14 6 2/6 E Schoklitsch 1934 b 240 4140 0.22 3 5/5 E Einstein, bedload b 240 6120 0.33 3 6/6 E Sand-bedded, large Toffaleti, total 0120 0242 0.08 6 5110 A Einstein, modified 0\00 011 0 0 5 6/6 A EngelundiHansen 080 1 140 0.07 5 6/6 A Inglis/Lacey 080 1 \0 0 0.09 4 4/4 A Colby 1964 040 280 0.20 2 3/4 C Ackers/White t 280 4 130 0.24 5 2/6 0 Bagnold 1956 b 240 350 0.38 3 3/3 E Chapter 4: Channel transport and storage 101

Table 11 : (continued) Equation type: t = total-load equation, b = bedload equation Table 10 histogram: distribution of the modes from rivers in Table 9; central category is within a factor of two of measured value Table 9 histogram: sum of histograms for the given river class in Table 9; central category is within a factor of two of the measured value Table 9 ratio: proportion of histogram entries from preceding column that are not in the central category Rivers: number ofrivers tested within the river category Independent tests/total: number of comparisons by independent users for river class / total number of comparisons Rank: listed from A (high) to E (low) according to the following criteria: A Three or more channels tested o Two or more channels tested Modes for each river tested within a factor Modes for each river within a factor of2 of of2 of measured values for all (Table 10) measured values for at least 75% of tests Fewer than 10% of test quartiles off by more Fewer than 50% of test quartiles off by more than a factor of2 for all tests (Table 9) than 2x for all tests (Table 9) Three or more independent tests At least one independent test

B Three or more channels tested E Two or more channels tested Modes for each river within a factor of2 of Modes within a factor of 2 for more than measured values for all tests (Table 10) 50% of all tested Fewer than 25% of test quartiles off by more Fewer than 50% of test quartiles off by more than 2x for all tests (Table 9) than 2x for all tests (Table 9) Two or more independent tests At least one independent test

C Two or more channels tested Modes within a factor of 2 for all tested Fewer than 25% of test quartiles off by more than 2x for all tests (Table 9) Two or more independent tests

average. An additional complication arises because some publications compare total-load calculations to measurements of bedload or vice versa. We took these differences into account to compare the utility of different equa tions. Comparisons between predicted and measured values are described in Tables 9, 10, and 11 by three-digit numbers that represent three-celled histograms. The first digit of each number is an index of the number of underestimated cases, the central digit indicates the cases for which the modal estimate is within a factor of 2 of the measured transport, and the third digit in.

Table 12: Characteristics ofstreams used to test sediment transport formulas A. GRAVEL-BEDDED CHANNELS

Sediment 0 50 River Discharge Depth Width Slope bed armor Method* code name (m J/s) (m) (m) x 0.001 (mm) (mm) Q, Qb AA Aare 44-125 1.3-2.4 15-26 2.3 48-68 70 spl BG Bas de Glacier 4-12 70 60-80 spl Arolla CP Cache La 0.3-21 6-19 10-20 33-42 Poudre CW Clearwater 290-3510 3-7 130-150 0.1-0.7 18 72 spl EF East Fork 2-42 0.6-1.7 15 0.7 1 spl EL Elbow 35-109 0.5-1.1 39-47 7.4 27 76 spl GO Goodwin 0.9 8 10 spl spl JO Jordan 56-200 1-2 19-43 30-35 80-300 KA Kawerong 4-6 20-71 6 ME Meshushim 200-300 2 26-31 30 200-300 OA Oak Creek 1-3 0.3-0.5 5-6 10-11 20 54 spl OH Ohau 450 20 75-100 mmt RA Rakaia 1250 gravel SB Sacramentol 150-2240 2-7 144-160 0.1-0.3 0.4-6 spl Butte SC Sacramentol 140-1120 3-8 80-113 0.1-0.2 0.2-2 spl Colusa SL Slate 1-3 7 26 43 SN Snake 2200-3500 4-5 165-192 0.7-1.1 32 54 spl TA Tanana 410-1680 2-3 107-466 0.5-0.6 8 12 spl TW Tanllwyth 0.5-1.4 3 40 81 VE Vedder 216-370 1-2 85-90 2.0 19 44 WA Waimakariri 90 100 4.8 28 mmt *Method: spl direct measurement of quantity mmt surrogate measurement used, such as bedform velocity (Q, = suspended bed material load, Qb = bedload)

or under-prediction by the equation, and results are adjusted accordingly. For example, most data show that roughly half the bed-material load is carried in suspension in sand-bedded rivers. Results are summed for different comparisons on the same river, and the result ing histograms are shown in Table 9. Table 9 may be used to: 1) determine whether a particular equation is likely to under- or over-estimate values for a particular type of river; 2) identify which equation performs best for a particular channel; and 3) determine how consistent the results are for the channel. Rysults of the comparisons are summarized by channel type in Table 10. Table 11 then ranks each equation according to the number of channels for which it produces valid results within each category and how consistently valid those results are; derivation of the rankings is described in the notes accompanying the table. Chapter 4: Channel transport and storage 103

Table 12 - (continued) B. SAND-BEDDED CHANNELS

River Velocity Depth Width Slope Bed D50 Method* : code name (m /s) (m) (m) x 0.001 (mm) Q, Qb Q, AT Atchafalaya 1.7-2.0 14-IS 0.05 0.09-0.12 spl mE CO Colorado 0.8-1.0 1-4 107 0.1-0.3 0.4 spl mE FI Fish Creek 0.2-0.S 0.3-1.2 0.08-0.1 0.4 spl =Qb ID Idle 0.3-0.6 0.S-1.3 0.3 tr tr JC Japanese chs. 0.6-0.9 0.2-0.7 1.3-1.4 ML Middle Loup 0.6-1.1 0.2-0.6 23-47 0.8-2 0.1-0.3 spl spl MN Mountain Cr. 0.5-0.8 0.1-0.4 4 1.6 0.9 spl MO Missouri at 0.9-2.7 2-6.6 300 0.15 sand spl mm Omaha MS Mississippi at 1.5-2.4 9-17 450-540 0.01-0.1 0.3 spl mE St. Louis MT Mississippi at 1.6 IS -16 0.3-1 sand spl mE Tarb. MU Muddy Cr. 0.3-0.8 0.1-0.5 4-5 0.6 spl spl MY Money Cr. 0.7-1.5 0.6-1.7 7-12 0.9 0.5 fill NI Niobrara 0.6-1.0 0.3-0.5 21-41 1-2 0.1-0.3 spl spl PA Paraguay 0.S-0.9 6-12 0.2 spl mmt RE Red River 0.8-1.1 6-7 0.07-0.08 sand spl mE RG Rio Grande 81-197 0.2-0.5 spl mE RP Rio Puerco 0.4-2.2 0.3-0.8 1.1 0.2-0.3 spl SJ San Juan 0.7 0.8 30 0.6 0.3 spl SK Skive-Karup 0.6 1.0 8 0.5 spl WG W. Goose Cr. 0.4-1.4 0.04-0.08 4 2.5-3.2 0.3 spl ZA Zaire 11-13 1300 0.06 0.6 spl *Method: spl direct measurement of quantity tr sediment tracers used mE modified Einstein method used to calculate portion of load mmt surrogate measurement used, such as bedform velocity fill infill measured in reservoir or depositional site (Qs = suspended bed material load, Qb = bedload, Qt = total bed material load)

Equations have been most comprehensively tested for sand-bedded channels. In small sand channels (widths less than about 20 m), the equation by Ackers and White (1973) provides the best results for total bed-material load. In moderate-size channels, the Engelund-Hansen equation (described by Vanoni 1975) produces consistently accurate results for total load and can also be used to calculate bedload. The Toffaleti (1969) equation shows the best results for large channels but requires measurements of suspended load, while the Engelund-Hansen equation produces acceptable results for cases where these measurements are not available. The Engelund-Hansen, Ackers-White, and Toffaleti equations were each tested in more than a dozen channels, and each provided valid results for over 90% of those channels. Tests of bedload equations on sand-bedded channels are fewer and 104 Chapter 4: Channel transport and storage

Table 13: Studies that compare transport equations with data/rom natural channels. River codes are shown in Table 12, and equation codes in Table 8.

Reference Rivers Equations used Method Ackers & White 1973 IOPA AW d: list Amin & Murphy 1981 FI MMTo d: graph Andrews 1981 MU AW EH SH7 Y73 d: graph Bagnold 1966 CO RG RP SJ B66 d: list Bagnold 1980 CW EF EL JO ME SJ T A ZA B80 d: list Bathurst et al. 1987 AA CP CW EL OA SL SN T A TW S4 d: curve Bishop et al. 1965 COML NI RG BSR d: curve Bondurant 1958 MO La d: graph Carson & Griffiths 1987 OHRA WA B80 t Carson & Griffiths 1989 WA MM t Colby 1964 MORP C64 d: graph Colby & Hembree 1955 NI Ou Em Et S34 d: list Oiplas 1987 OA Oi d: graph Gomez & Church 1989 EF EL SN TA AW AWO AWS B80 Ou Eb MP d: graph MM P82 S3 S4 Ya Hollingshead 1971 EL Eb EmMM d: curve Hubbell & Matejka 1959 ML Ou Em Et Ka MM S3 d: list Kuhnle et al. 1989 GO MM P82 Y84 d: li st Laursen 1958 NIMNWG La d: curve Maddock 1983 NI ML MN MS RG Ma r: % in rg Mahmood 1980 MO AW C64 Eb Em Et La MM To d: list Nakato 1990 SB SC AW EB EF EH IL MM Rb Rt S3 d: graph To Y73 Parker et al. 1982 CW ELOASN VE Et MM P82 d: graph Pickup & Higgins 1979 KA Mad MM Sh Y73 Rijn 1984a JC MN SK AW EH MM Rb r: % in rg Rijn I 984b MLNI AW EH Rs Y73 r: % in rg Shen & Hung 1972 AT ML MS MTNI RE SH7 d: curve Shen & Hung 1983 ATMS MTRE Em SH8 d: list Stall et. al 1958 MY Ou Eb S3 Stevens & Yang 1989 MNNI AW C64 Eb EH Et Ka La MM Ro r: mean S3 To Y73 Thompson 1985 OH AWEB MM c Toffaleti 1969 AT ML MS MTNI RA RE RF To d: curve Vanoni et al. 1961 COMNNI WG Ou Eb MP MM S3 Sh d: curve Vanoni 1975 CONI BI C64 EB EH IL La To d: curve Warburton 1990 BG S4 t White et al. 1978 AA AT EL ML MN MS MTNI PA AW B56 B66 BI BSR Eb EB EH r: mean SK WG Et Gr IL Ka La MM Ro Sh To Ya Yang 1973 MLN1 Y73 d: graph Yang 1979 ML MN MS NI RA RF Y73 Y79 r,d: curve Yang & Molinas 1982 ML MN MS NI RA RF A W C64 EH Ma SH7 Y73 Y79 d: graph Yang & Stall 1976 ML MN MS NI RA RF B66 C64 Ou La Em Y73 d: curve * Method used to compare predictions to measurements: c comparison of predicted curve and rating curve; no data are presented d data are listed or graphed, or predicted curve is compared with measured points r ratios of predicted to measured load are averaged or tabulated by range ratio between sum of predicted values and sum of measurements Chapter 4: Channel transport and storage 105 generally show poorer agreement. However, both the Toffaleti and Engelund Hansen equations can be used to calculate the bedload component in isolation, and their excellent record in calculating total load suggests that the bedload components are likely to be relatively accurate. Less information is available for gravel-bedded channels, and very few equations were found to perform consistently well in any category. The Bagnold (1980) equation provides consistent results for medium-sized channels, while the Meyer Peter and Miiller equation works well for large, braided rivers. Other equations accurately predict loads in particular channels, but comparisons are too few to allow general recommendations. Few equations are appropriate for gravel channels with paved or armored sur faces. Equations developed by Parker et al. (1982), Diplas (1987), and Parker (1990) are designed specifically to account for armored beds, but they have rarely been tested and none includes the magnitude of armoring as an explicit parameter. Results for these equations thus are likely to be valid only for channels armored to the same extent as Oak Creek, where all were calibrated. Bagnold's (1980) equation appears to provide adequate predictions for medium-sized channels with coarsened surfaces, probably because it is essentially a statistical summary based on bedload measurements from a number of channels in different regions. In general, total-load equations or those constructed using experimental results from sediment finer than 15 mm tend to overestimate transport rates for paved gravel beds. The data sets used to calibrate these equations generally did not include armored conditions, so the subsurface grain-size distribution is implicitly assumed to be representative of the surface grain size, and the moderating effect of the pavement thus is not accounted for. In contrast, bedload equations and others developed using at least some sedi ment coarser than 15 mm are more likely to provide valid estimates for armored conditions. It should be noted that many of the measured transport rates used for comparison are not particularly accurate. Bedload is difficult to measure, and many samplers are not capable of distinguishing between sediment suspended near the bed and actual bedload. In addition, transport processes are rarely continuous. Bedload often moves in sheets or dunes, for which periods of high sediment discharge are fol lowed by periods of inactivity at a site (Gomez et al. 1989). Carey (1985) found that rates of bedload transport varied from 0 to 4 times the mean rate during a 4-day period of uniform discharge. Rates may also vary widely through time due to changes in the structure of the bed surface (Reid et al. 1985, Gomez 1983). Lack of agreement between measured and calculated rates thus may not be the result of an invalid equation. In compiling the data it was noted that different investigators often calculate different values using the same equation on the same data set. Some of these differences arise because a user selected a subset of the available data from a river. 106 Chapter 4: Channel transport and storage

In other cases, different methods are used to differentiate the washload component or to estimate the unmeasured component of the suspended bed-material load. In still other cases, differences have probably resulted from arithmetic errors or use of an invalid version of an equation; many published reviews and even some of the source documents contain grave errors in notation. The discrepancies in published results indicate that application of transport formulas is not a routine task, but requires considerable understanding of sediment transport processes and the attrib utes of particular equations to decide how · the equations' parameters are to be defined in a field setting.

4.6.3 Applying sediment transport equations A qualitative understanding of sediment transport at a project site will greatly improve the utility of transport calculations. A first step in applying a transport equation thus consists of qualitatively characterizing the channel, as discussed in a previous section. The major sources of sediment and the sediment-transporting flows should be identified, as well as the downstream pattern of channel erosion and deposition. On this basis, an appropriate site can be selected for calculations. Ideally, the site will be along an unobstructed alluvial reach where the channel is straight, of uniform gradient, and has a homogeneous bed surface. Many gravel-bedded channels have coarse surface layers which shield finer subsurface sediment from transport until flows become high enough to mobilize part of the surface layer. Some channels transport finer sediments atop the coarse pavement at low flows, and this component is referred to as the through-put load. This condition must be identified in the field to prevent fatuous transport calcula tions. Channels with through-put loads are usually evident from the presence of a ribbon of sand atop a continuous gravel pavement. Once recognized, through-put sediments are excluded from transport calculations for the pavement and subsurface bed material. If the through-put component is considered important, transport calculations can be carried out independently for this fraction by using the width and texture of the transport band and the grain-size distribution of the through-put load in calculations. Measurements at the evaluation site will include the cross-sectional geometry, the water-surface gradient at the discharges of interest, and the grain-size distribu tion of the through-put, pavement, and subsurface layers. Water surface gradient should be averaged over at least one meander wavelength, or over a distance of about 10 times the the channel width (figure 17). The extent of the active transport zone should be determined, and grain sizes must be tabulated separately for active and non-active zones. Sites usually preserve evidence of the effects of past flows, and the largest grain sizes moved during the last flood should be noted if possible, as well as the high-water mark from that event. This information can be useful for testing the validity of an equation's predictions. The minimum grain sizes of immo bile sediment should also be noted. Chapter 4: Channel transport and storage 107

Box 21 Defining the zone of active transport After a transport event, gravel will be relatively loose; you are likely to sink in slightly as you walk across the zone, and clasts will shift. Particles may be imbricated and aligned strongly by the flow Particles tend to have clean surfaces, and some crystals are shiny Sediment bars have little, if any, vegetation The active zone tends to be in the deeper part of the channel and across the up stream shoulders of bars Grain size may be smaller than in adjacent zones

If the grain-size distribution of the bed surface is not the same as that of the subsurface bed material, then calculations will need to take into account the pres ence of the pavement. Transport equations are likely to be useful only if applied to discharges higher than the threshold for mobilizing the bed surface, since the grain diameter used in most equations reflects the size distribution of the subsurface bed material. The threshold is most confidently identified if observations of bedload transporting flows have been made. In the absence of such observations, the equa tions for initiation of motion presented in a previous section may be used to esti mate values. Many bedload formulas contain a critical shear stress for each particle size or for the average grain size, and the equations are very sensitive to variation of this parameter. The uncertainties in estimating this critical stress (described in an earlier section) therefore seriously degrade efforts to predict transport rates, and make worthwhile the expenditure of some effort to constrain these rates using local field evidence. Once field measurements have been made, characteristics of the measurement site and design discharge can be compared to the channels described in Table 12, and Table 9 then can be used to identify bedload or total load equations that have provided accurate estimates for similar channels. In some cases, the type of data available for the measurement section may constrain the choice of equations. Because of the inherent imprecision of bedload calculations, it is wise to calcu late rates using as many equations as possible to assess the accuracy of the ,esti mates. It is likely that some of the untested equations will eventually be shown to be useful for particular channel types, and if the conditions for which an equation was developed match those of a project site, then the equation might justifiably be applied for comparison to predictions of more carefully verified equations. Other equations may be optimal for particular channel conditions even if they do not perform well for other channels of the same size. Thus, if a project channel is similar to a particular test channel (table 12), then any equation that has proved accurate for that channel may reasonably be used. 108 Chapter 4: Channel transport and storage

Extreme care must be taken in applying the selected transport equations. Carson and Griffiths (1987, pp. 61-80) illustrate the types of interpretations that are in volved in successful use of bedload equations for field conditions, and Parker et al. (1982) demonstrate the importance of selecting a valid grain-size parameter when standard equations are applied to armored or paved gravel channels. Both of these studies illustrate how important it is to understand the factors controlling bedload transport before attempting to use an equation. F or example, calculations using transport equations cannot be applied to cross stream averages in channels with even moderate cross-stream variations in depth. Flume experiments generate transport that is relatively uniform across a channel, while natural channels experience wide variations in transport conditions across a cross section. Valid computations must therefore take into account cross-stream variations in particle size, flow depth, bedforms, and velocities, and only in flume like channels will average parameters for a channel cross section produce valid results. Any sediment transport computation for a non-uniform channel cross section should be applied to increments of width and the results integrated across the channel. These increments are most useful if they take into account the bounda ries of active transport zones. If such an approach is not taken, transport calculations may severely underestimate transport rates where the transport zone is restricted to the deepest part of the channel. Care must also be taken in defining the parameters to be used in the equations. To determine which grain-size parameter should be used in a transport calculation, it is necessary to identify the portion of the bed material that constitutes bedload at various flows. Pavement is usually absent or is weakly developed if the bed material is fine-grained and frequently mobilized, and the problem of determining the bedload size fraction is then a matter of separating out the suspendible fraction at each flow (see later). In channels with gravel or coarser beds, however, recognition of the bedload fraction is more complicated, and methods for doing so are still being investigated. Few generalizations have yet emerged from this work, but several studies suggest tentative guidelines for selecting particle sizes for use in transport equations. Research by Parker and Klingeman (1982) and Dietrich et al. (1989) on the initia tion of transport and the development of pavements suggests that almost all size fradions in the bed material are equally mobile over the long term, so the texture of the time-integrated bed-material load should approximately equal that of the sub pavement layer. Thus, the sub-pavement Dso should be the appropriate value to include in bedload or other gravel-based equations, and if a pavement exists, it should be removed and the underlying material sieved. This may present a consid erable sampling problem, although the degree of difficulty is not yet known because there have been few studies of the variability and spatial pattern of sub-pavement particle size. The problem is reduced where no pavement has developed and in straight reaches with only small variations of texture. On. the other hand, the sub- Chapter 4: Channel transport and storage 109 pavement Dso is likely to produce an overestimate of transport rates if a total-load or other sand-based equation is used, but these would not ordinarily be used for gravel bedded channels. In any case, calculations based on the surface grain-size distribu tion are likely to underestimate transport rates unless an equation is used that was developed using surface grain sizes in channels with a similar ratio of surface to sub-surface grain sizes to that present in the channel for which calculations are being made. A few channels produce bedload samples that are consistently finer than either the pavement or sub-pavement. !,here are several possible reasons for this: the trap efficiency of the bedload sampler for large particles may be low; the largest flows may be under-represented due to logistical problems or weather patterns; the channel bed may have recently been armored by changes in flow regime; or sedi ment may be contributed from two sources, as where sand from a tributary basin is transported at moderate flows over mainstream cobbles that move only during high flows. The downstream ends of point bars often exhibit a typical grain-size distri bution of bedload, either because pavements are absent at those locations or because near-bed flows deposit the most frequently transported bedload there. Observations at these sites may provide an estimate of bedload size in relatively frequent flows, but such an estimate should be supported by additional data. Some interpretation is also necessary in selecting the shear stress value for use in equations. Where channels are straight, flume-like, and free of bedforms and obstructions, the shear stress available for sediment transport is approximately equal to the total basal shear stress. At more complicated sites, however, the two values can be very different. Carson and Griffiths (1987) emphasize the necessity of using a local shear stress rather than the total boundary shear stress where channels are obstructed by debris or form multiple threads. The Meyer-Peter and MOller equation explicitly accounts for the local boundary shear stress by using a grain-roughness to calculate the boundary roughness parameter for the flow, but even this equation subtracts only the roughness due to bedforms rather than that exerted by major obstructions or flow divergences. Many other equations were developed using data from flumes where the only effective roughness was that imparted by the bed sediment; in this situation use of total boundary shear stress (r = pghS) is a good approximation for that experienced by the bed. These equations should not be used in natural channels with roughness imparted by obstructions and bedforms, al though substitution of a measure of local boundary shear stress for the total bound ary shear stress specified by the equations might improve their predictions (Carson an~ Griffiths 1987). Sediment transport computations must be integrated over a flow-duration curve to calculate sediment discharge over a period of time. If the cross section is un gauged, how well the flow-duration curve can be estimated is an important con straint on the precision of the computed load. Duration curves are usually estimated by analogy to gauged sections on similar reaches in the region, but results are 110 Chapter 4: Channel transport and storage

Box 22 Checklist for applying sediment transport equations 1. Examine the channel system: How are the major sediment-transporting flows generated? Where are the alluvial reaches? Where are channels aggrading or incising? Is there much downstream sorting? Where are the sediment sources? Is there adequate sediment available for channel transport? How important are obstructions, dams, organic debris, and other complexities? 2. Select a site for calculations and measurements that: Is alluvial Is straight Has a uniform gradient Has no obstructions Has a homogeneous bed surface Carries flow in a single thread during sediment-transport events Is not near tributary confluences or major sediment sources 3. Examine the calculation site: Is there much at-a-site sorting? Is there a through-put load? Where is the active transport zone? What sizes are mobile here? Is the pavement strongly developed? Is the bed surface an immobile armor layer rather than a mobile pavement? What zones should the cross section be broken into for transport calculations? Does the reach show evidence of aggradation or incision? 4. Provisionally select a discharge for calculations 5. Make the necessary measurements: Cross-sectional geometry, including identification of the active transport zone Grain-size distribution of pavement in active and non-active zones Grain-size distribution of subsurface sediment in active and non-active zones (continued)

approximate at best. Because most bed-material transport in gravel channels occurs during a few high flows, high stages are the most important ones to define for sediment transport calculations. Carling (1988), for example, finds 'that 40% of the bedload transport in one alluvial channel occurs during discharges of between 80% and 100% of the bankfull discharge, and that another 40% occurs at still higher discharges. Chapter 4: Channel transport and storage 111

continued: Box 22 Water surface slope at the relevant discharges Maximum grain size transported during the last event Crest stage of the last transport event 6. Calculate the basal shear required to initiate transport in the bed zones and verify that the selected discharges will mobilize the surface sediment 7. Select appropriate transport equations based on: Whether the channel conditions conform to those used to derive the equation The size of sediment transported Whether bedload or total load is to be calculated The type of data available How well an equation has been shown to perform for channels of similar size The extent to which the equation has been validated" for similar channels 8. Check the equation for notational errors by comparing several sources 9. Calculate potential maximum loads for selected discharges using equations Calculate for width increments across the channel, rather than using cross stream averages Use as many equations as are likely to apply Pay particular attention to the appropriate grain-size parameter used by the equations 10. Evaluate the importance of channel conditions that may influence results Evaluate the effects of channel characteristics that do not conform to the ideals listed in step 2. Are there local flow accelerations that would make flow behave unlike that in a flume? Calculate the ratio between the bed surface and sub-surface median grain sizes to estimate how supply-limited the channel is likely to be. 11. Test the results Compare predictions of different equations Test sensitivity of results to your measurement confidence intervals Compare results to measured loads, measured rates of accumulation in channel reaches, or reservoir sedimentation data, if available

Transport formulas are constructed to calculate the potential sediment transport rate assuming an unlimited supply of sediment of the given grain-size distribution. Transport rates are thereby assumed to depend only on flow and sediment charac teristics. However, Mosley (1981) reports that bedload transport.rates in a small forested stream are most closely controlled by sediment supply: much sediment is 112 Chapter 4: Channel transport and storage transported immediately after woody debris jams are washed out, but little is carried at other times. In extremely supply-limited channels such as this, the total annual bed-material flux predicted by even the most appropriate formulas is usually far larger than measured values, and equations can be used only to provide crude estimates of spatial patterns of transport capacity. These channels can be recognized because sediment of transportable sizes is usually segregated into relatively small deposits along the stream or is present only where trapped by obstructions. In cases where the supply limitation is less extreme, the condition can often be recognized from observation of sediment-supply processes in the basin and exami nation of the grain-size distribution of bed material. Supply-limited channels usually exhibit few near-channel sources of bed-material-sized sediment, and channel bed surfaces usually are comprehensively paved by coarse sediment. Banks tend to be relatively stable, bedrock may be exposed along portions of the channel, and deposits of readily transportable grain sizes are sparse. In contrast, channels with abundant sediment supply tend to show little difference between the grain-size distributions of the bed-surface and subsurface layers. Dietrich et al. (1989) suggest that the ratio between the median grain diameters of the pavement and subsurface might be a useful indicator of the relative values of sediment supply and potential transport rates at a site. If the ratio approaches 1.0, then sediment transport is not expected to be supply limited, and transport calculations are more likely to be valid. Because the condition of sediment supply may be reflected by the extent of armor ing, a transport equation that explicitly accounted for the difference between surface and subsurface textures might eventually be developed that can be applied to supply-limited channels. A surprisingly high proportion of the literature on sediment transport equations contains algebraic, arithmetic, and notation errors. Not only have some review papers misquoted equations, but some of the original sources contain typological or conceptual errors that have since been rectified. It thus is useful to compare differ ent versions of the same equation before it is applied, and to check each equation for dimensional consistency. The reasonableness of calculated results should also be considered critically. Results of sediment transport calculations can be checked in several ways. Different equations incorporate different assumptions and use different flow pa rameters, so if they predict similar results, sediment transport in the channel being assessed probably is relatively uncomplicated. A higher level of confidence can then be placed on the predictions than if results are widely divergent. Calculations can also be carried out for the maximum likely under- and over-estimates of equa tion parameters to determine the sensitivity of results to measurement errors. If results are highly sensitive to small variations of a poorly defined parameter, then less faith will be placed in their accuracy. Wherever possible, predictions should be compared with field measurements. Chapter 4: Channel transport and storage 113

4.6.4 Use of field observations Even where channel morphology is simple enough that sediment transport equations can be used, transport predictions may be unreliable because of uncer tainties in flow frequency and particle size, shortcomings of the available formulas, limited sediment supply, and a site's failure to meet the equations' assumptions. Einstein and Chien (1955), for example, find that the Einstein (1950) equation is not appropriate for flows with high near-bed concentrations of suspended sediment, and other undiagnosed complications are as likely to beset other equations. Independent field evidence of sediment transport rates ~hould thus be used whenever possible to check the accuracy of predictions. Verification is most often done by surveying and dating the volume of sediment trapped in ponds, reservoirs, or aggrading channel reaches (Dunne 1988, p. 424), but other measurements are also possible. For example, Collins and Dunne (1989, and see later) compare predictions of annual bed-material flux by the Meyer-Peter-Muller and Parker methods with estimates based on a locally calibrated percentage of the measured suspended load, measure ments of bar accretion, and records of gravel mining and net channel scour. Maddock (in Vanoni 1975, p. 348) describes the expected proportions of bedload for several channel types (table 14). Here, "total load" also includes washload, in contrast to the definition used for sediment transport equations. Thus, predicted bedload transport in gravel-bedded streams is likely to approximate the total bed material load, but the bed-material load represents only a small fraction of the total sediment load. Much of the sand fraction is removed from these streams as washload and so is not represented in the bed material, and bedload is likely to represent less than 15% of the total sediment load. In contrast, only silt- and clay sized particles are removed as washload in many sand-bedded channels, and the bed-material load thus makes up a much higher proportion of the total sediment load. Even though half of the bed-material load may be traveling in suspension, the bedload fraction usually constitutes an important portion of the total load. Many sites have measurements of suspended load only, and the relations shown in Table 14 can be used to estimate bedload transport from the time-integrated suspended load. The U.S. Geological Survey publishes annual data summaries for sediment and flow data on rivers they gauge (e.g. USGS 1989), as does the Canadian Inland Waters Directorate (e.g. CIWD 1977). At a few ofthese stations a small number of bedload transport measurements have been made along with the more common suspended sediment measurements. These values can be used as local checks on the proportions of bedload given in Table 14. The river gauging network reached its maximum extent in the United States in 1982, and many stations have now been abandoned (Osterkamp and Parker 1991), so old records should be checked to see if analogous channels were gauged in the past. 114 Chapter 4: Channel transport and storage

Table 14 - Expected percentage of bedload in total load (modified from Maddock, reported in Vanoni 1975, p. 348), where total load here includes bedload, bed-material suspended load, and wash/oad. Note that this definition of total load differsfrom that used in sediment transport equa tions by the inclusion ofthe washload component. Concentration Bedload discharge, of suspended Type of material as percentage of load, parts per forming channel Texture of suspended total sediment million of stream material discharge less than 1000 sand similar to bed material 20-75 less than 1000 gravel, rock, or small amount of sand 5-11 consolidated clay less than 1000 gravel silt and sand 4-5* 1000-7500 sand similar to bed material 9-26 1000-7500 gravel, rock, or 25% sand or less 5-11 consolidated clay over 7500 sand similar to bed material 5-13 over 7500 gravel, rock, or 25% sand or less 2-7 consolidated clay

* based on our analysis of simultaneous bedload and suspended load samples at five USGS gauging stations in Washington

4.6.5 Evaluation of the wash load component It is sometimes useful to estimate the grain sizes that are suspendible at various flows to assess the proportion of the incoming sediment load that is flushed rapidly through a channel. The grain-size distributions of the sediment influx, obtained by an analysis of sediment production, can be compared with the grain-size distribution of the bed material beneath the pavement. For example, colluvium in a drainage basin may contain a large fraction of silt and fine sand, while the sub-pavement bed material contains nothing finer than medium sand. Sediments finer than medium sand are then presumed to be flushed through as suspended sediment. Such field observations can be used to test the validity of prediction formulas. If formulas are found that are accurate for those field conditions, they may then be used to general ize and extend the field observations. The suspendibility of a particle is usually defined by the condition

p < Ws (4.13) u. Chapter 4: Channel transport and storage 115 where Ws is the settling velocity of the particle, a function primarily of its grain size, and u* is the shear velocity of the flow, defined by:

u. = ir/p (4.14) and

't = pghS (4.15) P thus is dimensionless if Ws and u* are in the same units. Dietrich (1982) outlines a method for estimating the settling velocity of natural particles. The critical value of P is not yet well defined, but it could be checked approximately by observing the finest component of the bed material, and by collecting a few suspended sediment samples and measuring the largest suspended particles using methods outlined by Guy (1969) and Guy and Norman (1970). In the absence of good field data, meas urements reviewed by Komar (1980, p. 318) serve as a suitable first estimate. After comparing the available data, Komar chose to use a suspension criterion of P < 1.25, and estimated that a value of P < 0.05-0.10 would be appropriate for washload. However, all of the available experimental values have been obtained from pipes, laboratory flumes, or sand-bedded channels with relatively low gradi ents, so it would be wise to check the applicability of these P-values to small, rough, mountain channels through some simple field measurements. For example, sieving of bed material indicates the approximate upper size-limit of washload as the minimum size present in more than trace amounts, since only a small volume of washload is usually trapped in the bed. Measurement of the maximum grain size collected with a depth-integrating sampler or by grab sampling allows P-values to be obtained for the bed-material suspended load. Again, we cannot overemphasize the usefulness of a few local measurements collected within a general theoretical framework such as that given by Equation 4.13 . Suspended loads in steeper headwater channels tend to be determined more by the activity of erosional processes than by the transport capacity of the channel. Bathurst et al. (1986) find that sediment rating curves change markedly with sea sonal changes in sediment source for a river in the Colorado Rockies, and Van Sickle and Beschta (1983) are able to improve predictions of suspended sediment transport by taking into account the volume of sediment available for transport. These observations lend support to the useful assumption that a high proportion of the suspendible sediment introduced to small streams is carried out of the basin. This assumption allows suspended load to be estimated crudely on the basis of sediment input as defined by a sediment budget. We will not review here the methods for predicting or monitoring suspended sediment transport, as they are described in various manuals (e.g. Vanoni 1975). 116 Chapter 4: Channel transport and storage

4.6.6 Limits of sediment transport predictions It has been demonstrated above and will again be shown in Chapter 5 that even when sediment transport equations are used with extreme care, they provide only a crude estimate of the quantities of sediment transported and deposited, although their predictions of spatial patterns of transport along a valley or temporal changes due to altered flow regime are often accurate. In forested, mountainous terrain one frequently encounters channels where the available sediment transport equations are entirely inappropriate, and there is no prospect for future development of an appro priate equation. Such channels are often characterized by step-pool bed topography created by concentrations of large boulders or organic debris dams. Under such conditions, one can do little more than make empirical estimates of the past and current rates of accumulation behind the barriers and of sediment transport competence and rate in the first tractable reach downstream of the barrier strewn reach. One can also compare rates of input to storage with a sediment-budget estimate of the production rate for each grain size to estimate washload transport and bed-material flux. This information would also allow calculation of the sedi ment flux that would result if the organic dams were scoured away by a debris flow or flood, since the scoured channel would probably not provide opportunities for extensive deposition until the dams were re-established. This assumption could in, tum be tested by applying sediment transport equations to a scoured channel geometry. Examination of the age-distribution of organic dams in the region would indicate the time needed to re-establish the sediment storage capacity after the channel is scoured. Thus, even where standard sediment transport equations are not useful, one can provide much useful quantitative information about the sediment transport system by combining local measurements of channel characteristics with a hillslope sedi ment budget. On the other hand, routine calculations of sediment transport based largely on office work without a thorough field interpretation of transport condi tions are not likely to produce useful results.

4.7 Sediment storage Sediment accumulates in, and is released from, channels and valley floors over periods that range from days to millennia. Such accumulations may affect the interpretation and documentation of sediment budgets for resource management. For example, Trimble (1981, 1983) shows that sediment held in long-term alluvial storage is now increasing sediment yields in a basin in Wisconsin, and Figure 24 (see later) implies a similar effect from meadow entrenchment in the Sierra Nevada. Church and Slaymaker (1989) demonstrate that current sediment yields in parts of British Columbia are dominated by remobilization of Pleistocene deposits along valleys, and that the sediment yields per unit drainage area thus increase down- Chapter 4: Channel transport and storage 117 stream. In other cases, rapid erosion rates on hillslopes are moderated by temporary storage of the eroded sediment before it enters high-order channels. Dietrich and Dunne (1978) calculated average residence times of 30 to 90 years in debris-flow fans at a site in coastal Oregon. It is not always necessary to document the effects of sediment storage. If the relative importance of different sediment sources is the major concern in basins without major potential for alluvial storage (such as in steep, narrow valleys), then one need not look beyond the hillslopes. In other cases, qualitative field observa tions or mapping of depositional forms and textures may suffice to assess the influence of storage in the sediment budget. For example, it may be important only to appreciate that none, or all, or an important fraction of the sediment load in some grain-size class is being trapped within the watershed, or to identify the areas most likely to experience deposition. Dunne (1988, p. 424; and see later) uses this type of information to infer that the rate of gravel accretion on an alluvial fan at the foot of the Cascade Range equals the gravel discharge immediately upstream. In other cases, it may be necessary only to determine whether the volume of sediment stored in a watershed is increasing or decreasing.

4.7.1 Identifying storage elements in channels Channel-related sediment storage can conveniently be subdivided into categories based on how frequently the sediment moves. Dietrich and Dunne (1978, p. 201- 203) and Dietrich et al. (1982, p. 15-20) discuss theoretical aspects of sediment storage and estimate residence times for several types of sediment reservoirs, including debris fans, active channel sediment (including bars, down to some rarely attained maximum scour depth), and floodplain sediment, although they overlooked the possibility that sediment could accumulate over centuries in some low-order channels until removed by debris flows (Benda and Dunne 1987). Kelsey et al. (1987) extend this concept of residence time to define four sediment reservoirs with differing relative mobilities in Redwood Creek, California: active sediment is present in the active channel, and usually moves at least once every several years; semi-active sediment comprises bars susceptible to revegetation, and is moved by flows with recurrence intervals of 5 to 20 years; inactive sediment moves only during extreme flows and usually is well-vegetated between events; and stable alluvium occurs in valley deposits that are not accumulating under the present regime but may be susceptible to channel erosion. Storage elements usually can be discerned on the basis of vegetation age and location with respect to a channel, although. an extreme flow can reset vegetation ages throughout the sequence. Dietrich et al. (1982) and Kelsey et al. (1987) treat the transfer of sediment between storage elements as a stochastic process that can be evaluated using field data. They use transition probability matrices to organize information on the probability of transfer from one storage element to another, and White (1986) uses a computer spreadsheet program to simplify such calculations. 118 Chapter 4: Channel transport and storage

Small channels in forested drainage basins may accumulate infrequently trans ported and inactive sediment in debris jams. The degree of sediment . transport activity in these deposits usually depends on the size of the jam relative to the channel size. Maser et al. (1988, p. 66) illustrate the relative stability of logs of different sizes as a function of channel width by noting the mean size of logs in a variety of channels, and Likens and Bilby (1982) perform a similar analysis for channels in New England. Channels vary widely in their opportunities to store sediment. High-gradient channels and those constrained by bedrock banks often show little floodplain development. Sediment is efficiently transported through these reaches, although it may accumulate temporarily behind debris jams. In fact, many mountain channels would store little or no sediment without the trapping function of large organic debris, and their sediment storage is thus very sensitive to removal or accumulation of such barriers. Steep reaches without large organic debris are not likely to be affected much by changes in sediment supply or transport capacity. Lower-gradient reaches usually provide more opportunity for sediment storage, and these sites often have well-developed floodplains and terrace deposits. Fine sediment is usually stored as over-bank deposits on floodplains. Most active floodplains are gradually growing in height, although aggradation may be balanced by stream-bank erosion of the floodplain deposits. Small changes in sediment input or flow characteristics at these sites can alter the balance between accumulation and erosion of sediment. The presence of once-active fluvial sediments in what are now rarely mobilized and inactive storage elements implies that a channel is capable of altering its form to reflect changing transport conditions; channel morphology has changed in these locations. The channel reaches most susceptible to change often can be identified by noting the distribution of storage elements. These features generally are identifiable on aerial photographs because they support characteristic vegetation communities.

4.7.2 Defining trends in channel-related sediment storage Where inputs to storage approximately equal erosion from those accumulations, storage can usually be ignored in sediment budgets. For the more complete budgets used in geomorphological research, however, this factor may be important as a mechanism for altering the character of sediment in transport. Bradley (1970), for example, finds that weathered clasts are more easily abraded during transport than fresh pebbles. Reconnaissance-level budgets usually consider storage only if there is an obvious pattern of aggradation or erosion, because in these cases hillslope sediment production will not provide an adequate estimate of sediment yield. Quantitative documentation of sediment storage is usually needed only if a sedi ment budget is intended to provide estimates of long-term sediment yield or sedi mentation rates. A net change in storage can occur either as a natural consequence of landform Chapter 4: Channel transport and storage 119 evolution or as a response to more rapidly changing conditions in a drainage basin. Love (1986) illustrates the first type of change in a discussion of the effects of tectonic deformation on sediment storage along the Rio Puerco. Natural aggradation is common at mountain fronts, where channel gradients decrease abruptly on entering the lowland. Large, aggrading alluvial fans are common at such sites. Valleys also often exhibit aggradation downstream of large, long-term sediment sources such as glaciers. Where land-use changes alter sediment input or transport capacity in a drainage basin, channels commonly respond by changing the volume of sediment in storage through incision or aggradation. These changes are usually limited to alluvial reaches, though they may shift the boundary between alluvial and non-alluvia~ reaches. Such changes often accompany urbanization or dam construction. Care must be taken to distinguish long-term trends in storage volume from those that represent cyclic or seasonal changes. Changes in storage can also merely represent a temporary response to an infrequent large event, such as a wildfire or landslide. Laird and Harvey (1986) describe a pattern of aggradation in low-order tributaries and incision of higher-order channels immediately following a wildfire in Arizona. After hillslopes recovered, low-order channels incised and larger channels. aggraded. Similarly, Sidle (1 988} documents a decrease in active-sediment storage during high flows in an Alaskan stream, followed by increasing volumes of active sediment in storage during subsequent low-flow years. Perkins (1989) shows that erosion of some landslide deposits in stream channels is inhibited by log jams at the front of the deposit, and that erosion of the deposits follows a predictable sequence (figure 23). Little or no erosion occurs during Phase I, which may be brief where organic debris is absent. This phase ends when the stream cuts through or around the deposit. In Phase II, rapid erosion occurs as the base level is lowered, and a new channel forms through the landslide deposit and the fluvial sediment trapped behind it. During this phase, 20-80% of the initial volume at Perkins' study sites was removed in less than 7 years. Erosion slows markedly in Phase III as a relatively stable channel geometry develops. Perkins shows that a diffusion model for erosion during Phases II and III can be calibrated by field surveys, so results of a few surveys may be sufficient to estimate the timing of sediment release from many organic-debris dams. Storage volumes can also change in a predictable seasonal pattern where sedi ment input and transport processes are non-synchronous. Duijsings (1986) notes that most sediment production in a Luxembourg catchment occurs during winter, and most of the contributed sediment is stored in channels until excavating flows occur during summer storms. McGuinness et al. (1971) find a similar pattern in the Coshocton basin in Ohio: during their study, most sediment entered the channel system from sheetwash during March, and sediment yields in 0.4-0.8 ha basins were highest then. In contrast, the highest sediment yields for the entire 15540-km2 basin 120 Chapter 4: Channel transport and storage

landslide deposited in stream channel J PHASE I 100 \V no erosion

PHASE II rapid erosion

PHASE III slow erosion small deposit, few logs, varies with size of initial deposit, t large stream valley shape W a 1------~------~ a Time

Figure 23: Phases in the erosion of landslide deposits or other sediment stored behind dams of organic debris. Source: Perkins (1989).

occurred during December floods. Bergstrom (1982) documents cyclic changes in the timing of incision and aggradation along channels at a badland site. Matherne and Prestegaard (1988) show that sediment delivery from a Pennsylvania watershed requires occasional high flows from summer thunderstorms to flush the sediments that have accumulated during snowmelt runoff. Several approaches can be used to identify and quantify trends in storage vol umes, irrespective of the causes for change. Areas of aggradation usually exhibit features that are obviously inundated by sediment. Haible (1980) found buried fence-posts in an aggrading fan, while Everitt (1968) and Sigafoos (1964) use evidence from partially buried riparian trees to demonstrate aggradation. Deter mining the ages of affected vegetation and structures provides the information necessary to estimate rates of change. Hupp and Bazemore (1991) use trees of a range of ages to discern temporal variations in aggradation rate over a long period. Aggradation rates may also be estimated by dating cultural artifacts buried in the deposits or by tracking inputs from a single event. Miller and Wells (1986) map the distribution of storage in a channel on the basis of the chemical characteristics of sediment and flow introduced by failure of a uranium tailings dam. Bed material may accumulate in bars that are later removed by large floods. Visual observations over a number of years, anecdotes from local residents, de- Chapter 4: Channel transport and storage 121

Box 23 Evidence for channel aggradation Heightened riparian water tables may kill some plant species, and non-riparian plant species may find themselves anomalously close to a stream The stream may display an anomalous channel geometry: the cross-sectional area may be too small to convey the expected bank-full discharge Overflow channels may erode across floodplains Surface debris, such as litter or organic matter, may become buried by new deposits Plants may be partially buried by new sediment Trees and shrubs may sprout adventitious roots along buried portions of their stems Bridge piers, roads, and fence lines may be partially buried Records at gauging stations will show changes in channel form Cross-sectional survey records at bridge locations can be compared with present cross sections Anglers usually are aware of changes in channel depth and form at their customary fishing spots Residents will be aware ofchannel changes at customary swimming holes Whitewater paddlers often can precisely describe changes in rapids Flood frequencies may increase Bank erosion rates may increase, widening channels and undermining vegetation The channel may radically change its course or begin to migrate more rapidly scriptions from kayilkers and anglers, sequences of aerial photographs, or records of bed elevation at stream-gauging stations may be used to recognize this kind of fluctuation. In addition, records of channel cross-section surveys are stored in the files of many engineering agencies, and artificial sediment extraction rates can often be reconstructed from dredging schedules and gravel-mining records. Collins and Dunne (1990) discuss methods for combining gravel-mining records with existing records of channel-bed elevation, aerial photography, and other evidence to docu ment changes in channel sediment storage. Rapid in-channel aggradation often increases the frequency and intensity of overbank flooding, destabilizes channel banks, and undermines riparian vegetation. Comparison of sequential photographs discloses the channel reaches that have widened during the photo interval. Grant (1988) explains how to use sequential aerial photographs to map the locations of channel aggradation along forested streams, and Ryan and Grant (1991) apply the method to evaluate the effects of logging and road construction on a coastal Oregon channel. Estimates of sediment loss to storage may also be made by comparing measured sediment discharge at several points along a river. Walling et al. (1986) use this approach to demonstrate that 28% of the suspended sediment entering a 13-km 122 Chapter 4: Channel transport and storage

reach of the River Culm in England is deposited there. A difficulty with this technique in many places, however, is that one is searching for relatively small differences between two large numbers (sediment yields); which are themselves poorly defined from inadequate records. Rigorous analysis of the sediment fluxes is therefore required. Sediment is often stored behind organic debris dams, and this complicates analysis of sediment storage in small forested basins (Keller and Swanson 1979, Swanson et al. 1982). Likens and Bilby (1982) propose a theory of the frequency and locations of such dams that relates dam stability to the relative sizes of the debris and the channels. Pearce and Watson (1983) carefully document sediment storage and release from deposits left by two episodes of landsliding in forested basins in New Zealand. Logs in the landslide debris are important for regulating the evacuation of sediment from the basins. Some sediment may also be detained in natural lakes or artificial impoundments. Large water bodies trap all but the finest sediment, although some flood-control dams are designed to flush sediment along the reservoir bed. The trap efficiency of large dams has been documented by detailed studies, and Brune (1953) describes a method for calculating trap efficiency. Griffen (1979) reviews methods for deter mining reservoir trap efficiencies in large reservoirs. Trap efficiencies in small impoundments are more difficult to estimate because conditions are more variable and efficiencies are more likely to change through time. Heinemann (1981) outlines a method for predicting trap efficiency in small reservoirs, and Moglen and McCuen (1988) discuss trap efficiency of small detention basins. Dendy (1974, 1982) measures trap efficiencies and sediment distribution in reservoirs with capacities of 0.88 to 117 ha-m in 0.75 to 35 km2 catchments and finds that all have average trap efficiencies of 80-95%. A smaller reservoir (1.1 ha area) has an average trap effi ciency of 77%, but monthly values range from 9 to 100% and may reflect differ ences in flow detention times through the year (Dendy and Cooper 1984). For individual storms, the average residence time of water in a reservoir (the impounded volume divided by the throughflow rate) may be compared with the time required for particles to settle through the water column to estimate the grain sizes that will be trapped. If the impoundment lies in a narrow valley, the average flow velocity (u) through it can be computed from the reservoir length and average residence time, and thus the average bed shear stress (t) can be calculated from p u2j 1:=-- (4.16) 8 where j is the Darcy-Weisbach friction factor (Leopold et al. 1964, p. 160) and p is the fluid density. The effect of widening, constriction, or other irregularities that modify flow velocity can be roughly evaluated and used to modifY the estimate of"C. The shear stress can then be incorporated into equations for initiation of motion Chapter 4: Channel transport and storage 123

(table 7), partitioning of transport mode (Equation 4.13a), and transport rates (table 8) to estimate which grain sizes will be transported through the impoundment as suspended load or bedload. Such calculations are very crude, so field observations at the project site or at comparable impoundments are vital for testing the applica bility and precision of calculations. Several studies have used bedload equations to predict downstream changes in sediment storage after dam construction. Flows are depleted of sediment at these sites because of deposition behind the dam, and the transport deficit is made up by entrainment of mobile particles until none are left. Pemberton (1975) uses the Schoklitsch, Haywood, and Meyer-Peter and Muller equations to predict removal of sands after closure of Glen Canyon Dam. Predictions were later verified by meas urements downstream of the dam. Hales et al. (1970) present a method for calcu lating downstream degradation using the Bagnold transport equation. Channel incision and excavation from storage is usually evident because of the presence of steep, eroding banks. In some cases the extent of incision or enlarge ment is indicated by exposed roots (e.g. LaMarche 1966, Germanoski and Ritter 1987), which also allow rates of enlargement to be estimated. Where changes are more subtle, they may sometimes be identified by comparing a channel's character to that of others in a region. Channels that drain areas of similar topography, bedrock, and vegetation in an area of uniform climate usually show similar relation shi ps between channel characteristics, such as width, depth, or channel capacity, and basin characteristics, such as drainage area. By plotting these relationships for channels in a region, those channels that do not fit the regional hydraulic geometry can be identified (Madej 1982, p. 101). Channels that show uncharacteristically large cross-sectional areas for a given drainage area are likely to have undergone recent enlargement and removal of sediment from storage. Similarly, channel enlargement or aggradation might be inferred if an estimate of bankfull discharge made using channel characteristics indicates a bankfull flow that has a very large recurrence interval for a basin of that size. Where gauging-station records are available, plotting gauge heights of low flows (as an approximation for bed eleva tion) against time can reveal trends of incision or aggradation (Collins and Dunne 1990).

4.8 Computations of sediment yield Occasionally it is possible to check sediment yield predictions by measuring the volume of sediment stored in ponds or reservoirs. To do so, it is necessary first to determine the proportion of the total sediment load that the deposits represent by estimating the trap efficiency of the water body (see previous section). Yields are usually presented in units of weight of sediment exported per unit time, so volumes of deposition must be translated to weight. Lara and Pemberton (1963) compile 124 Chapter 5: Applications measurements of unit weights of sediment of various characteristics and define regression equations to calculate bulk densities, and Carling and Reader (1982) report characteristic bulk densities of stream gravels. In many cases, sedimentation data already exist for reservoirs in the region of interest. Dendy and Champion (1978) compile reservoir sedimentation data from throughout the United States, and this information often can be updated by contacting the agencies that manage the relevant reservoirs. Reservoir sedimentation is often more useful than yields con structed from suspended sediment m~asurements because it generally represents a long-term record and it includes the bedload component. Van Sickle (1981) compares annual sediment yields measured by suspended sediment sampling in small Oregon catchments over periods ranging from 3 to 19 years and finds that they are approximately log-normally distributed. Because of this pattern, Van Sickle suggests that a median sediment yield is a better parameter for comparing watersheds than a mean y.ield, and further shows that more than 8 years of monitoring data are necessary to define the median yields to within a factor of two for most of the monitored catchments. Sediment may also accumulate at sites other than ponds. For example, Kreznor et al. (1990) are able to estimate the sediment yield since Euro-American settlement of an area in Illinois by measuring the volume of sediment overlying a buried A horizon containing fly ash on a valley floor.

5 EXAMPLES OF SEDIMENT BUDGET APPLICATIONS

Sediment budgets have been constructed for a variety of sites throughout the world (table 15), but the majority of those with published descriptions represent relatively large investments of time, and many include some monitoring. Results of such studies are useful for suggesting process magnitudes that might be expected at other sites, and they also demonstrate the variety of sediment production regimes possible (table 16). In addition, most studies incorporate some methods that are applicable to rapid budget evaluations or detail particular process interactions that should be evaluated in other studies. ·However, rapidly constructed budgets have rarely been described, and four short-term studies that we have conducted are describetl here to illustrate the approaches and techniques that can be used. All of the studies described were conducted to address questions related to land manage ment, and all were constructed over relatively short periods (see table 1). Chapter 5: Applications 125

Table 15 - Sediment budgeting studies Land use Cover Process Type Reference Site ADGILMRU CFGRVW ABCDEFGILPQRSTW OHRW Caine and Swanson 1989 CO · .. I ...... G .•. · .CD ...... QRS' .. .H.W Caine and Swanson 1989 OR · .. I .... .F .... · .CD. F .... QRST. .H.W Collins and Dunne 1989 WA . D ...... R •. A ... E .. 1. .Q.S .. · .R . Dietrich and Dunne 1978 OR · .. I .... . F ..•. A. CDEF .... Q . S .. ... W Dietrich et al. 1982 ...... · .C ....• L .. R ... OH .. Duijsings 1987 Lu .. . I .... .F ..•. · BCD .... L . QRS .. .H.W Gilbert 1917 CA ADGI.MR. CFG .. W A ..... G1.P.RS .. .H.W Kelsey 1980 CA · .G.L.R. . FG .. W AB ... F.IL ... S .. .H.W Kelsey et al. 1987 CA ...... R .. A ...... 1 .. Q .... · .R . Lehre 1982 CA · .. I .... . FG ... ABCD .. GIL . QRS .. .H.W Lehre et al. 1983 WA · .. I ...... v. AB .... G1. .QRS .. .H.W Leopold et al. 1966 NM · .G ...... W ABC ... G1.P.R ... .H.W Madej 1982 WA .. . IL.RU .F .... A.C.E.GIL.QRST . .H.W Madej 1987 CA ...... R .. A ...... 1 ...... · .R. McLean & Tassone 1991 BC . D ...... R .. A ...... !. .Q.S .. · .R. Megahan et al. 1986 ID ...... R. A ...... I.P .. S .. .. )W Parker 1988 NE ...... R ...... Q.S .. · .R . Ph ill ips 1990 NC A.G.L.RU . FG ... .B .... G .. P.RS .. ' .H.W Prestegaard 1988 PA A.G ...... FG ... .B .... G ..... S .. .H .. Reid 1990 Tz A.GI .. R. C.G .• W . B .... G .... R .. W .H .. Reid et al. 1981 WA ...... R. . F ...... FG.L .. R ... .H .. Roberts and Church 1986 BC .. . IL.R. . F .... ABC .. F .. L .. R. T . .H.W Sing 1986 CA ...... R •. AB ...... Q.S .. · .R. Smith and Swanson 1987 WA · .. I ...... v. A ...... L .. R ... .H.W Stott et al. 1986 Sc · .GI .... . FG ... . B ...... Q.S .. · .R . Sutherland & Bryan 1991 Ke · .G ...... • W ABCD ... I .. QRS .. .H.W Swanson et aJ. 1982 OR · .. I .... . F .... · . CD . F .. L . QRST . .H.W Trimble 1983 WI A.G ...... FG .•. AB .... GI.P.RS .. .H.W Location: Land use: Processes evaluated: Budget !YQe: BC British Columbia A agricu Iture A channel aggradation 0 overview CA California D gravel dredging B bank erosion H hillslopes CO Colorado G grazing C creep R river reach ID Idaho inactive D dissolution W watershed Ke Kenya L logging E transport abrasion Lu Luxembourg M mining F debris flows NC North Carolina R roads G gully erosion NE Nebraska U urbanized channel incision NM New Mexico L landslide OR Oregon Cover !YQe: P pond aggradation PA Pennsylvania C chaparral Q bedload Sc Scotland F forest R surface erosion Sw Sweden G grassland S suspended load Tz Tanzania R river T treethrow WA Washington V volcanic blast W wind WI Wisconsin W woodland tv -0\ Table 16: Results a/selected sediment budget studies.

Reference: Duijsings 1987 Reid 1981 Lehre 1982 Madej 1982 Reid 1989 this paper this paper

State or country Luxembourg WA CA WA CA WA CA County: Jefferson Marin Kitsap Contra Costa King Fresno Vegetation: forest forest grass/woods forest grass forest forest Land-use: natural natural grazing logging grazing logging logging Method: monitoring surrogate monitoring surrogate monitoring/sur- surrogate surrogate evidence of evidence of rogate evidence evidence of evidence of rates rates of rates rates rates Monitoring period (yr): 2 4 Area (km2) 10 1.7 38 0.11 155 6.8

Natural (tlkm 2) landslides 28 7 original 622 16 5 scar erosion 53 2 2 debris flows 10 treethrow 9 2 3 gullies 176 60 bank erosion 53% 29 16 6 8 burrows 4 2 sheetwash/splash 47% tunnels 9 {l Disturbed (tlkm 2) ,... landslides 22 '"..., roads 184 3 ~ surface 63 10 ;J:. roadcut ~ "- ditch 12 ;:;. \:l ~. 2) 16 0 Total (tlkm 80 867 110 89 193 ;::s "" Chapter 5: Applications 127

5.1 West slope of the Sierra Nevada, California Low-head hydro-power developments planned for six streams in the western Sierra Nevada would affect fish habitat, and planners wanted to know the type and magnitude of likely effects. They were particularly concerned about the extent to which the low dams and diversions would alter the availability of spawning gravels downstream of the projects. To estimate this impact, we needed to know: 1. The character and amount of sediment now in transport in streams, 2. The character and amount of sediment that would be introduced by construc tion and operation of the proposed developments, and 3. The effect of the proposed flow diversions on transport capacity and compe tence of the streams. Results accurate to within half of an order of magnitude would be sufficient. We first used aerial photographs and geologic maps to divide the six watersheds into regions of similar geology, soils, and topography, and we further subdivided these regions according to vegetation type and cover density. We visited represen tative sites in each category and found that the major sediment sources are bank erosion, treethrow, animal burrowing, dry ravel on roadcuts, sheetwash and rill erosion on disturbed and undisturbed slopes, and sheetwash erosion on road sur faces. We organized our understanding of the basin sediment transport systems using a flowchart (figure 24). This diagram is divided into three domains that represent different parts of the landscape: hillslopes, valley fill, and channels. Some hillslopes in the area abut streambanks, while others are estranged from the channel margins by valley-fill deposits. Hillslope erosion and transport processes are listed in the Hillslope domain, and arrows are drawn to indicate the fate of sediment transported by these processes. In some cases (e.g. animal burrowing and soil creep), all sediment is redeposited as colluvium on the hillslope or valley-fill deposits, while other processes deliver some (e.g. roadcut ravel) or all (e.g. riparian treethrow) of the mobilized sediment to channels. Bank erosion is the only process that contributes sediment to streams both from hillslope colluvium and valley-fill alluvium. If alluvial deposition and bank erosion were in balance for these fluvially deposited valley-fill sediments, bank erosion at these sites would not be included as part of the sediment production rate. In this case, however, entrenchment represents a recent change in sediment regime, and erosion of these sediments must be evalu ated. The structure of the flowchart showed what measurements were necessary to estimate the rate of sediment production: we needed to define only those rates indicated by arrows that cross the streambank. We then collected the information necessary to estimate the contribution from each of these processes. We used both field measurements and USLE calculations to evaluate sheetwash erosion. We measured rills and vegetation cover density 128 Chapter 5: Applications

HILLSLOPES

Colluvial storage (including valley fill) Hillslope sheetwash & rilling ancient streambank

Current streambank

CHANNELS

Figure 24: Flow chart illustrating the relationships between various processes of colluvium transport into small streams studied in the N. Fork Kings River basin, CA. Of the hillslope transport processes present, only riparian treethrow, bank erosion, hillslope sheetwash, road-surface erosion, and roadcut ravel supply sediment to channels. "Alluvial Valley Fill" here represents channel deposits of an earlier time; these are now being reactivated by bank erosion.

along slope transects, collected soil samples for textural analyses, and detennined the extent of annoring by point-counting bare surfaces on hillslopes, roads, and in the areas planned for temporary construction activities and long-tenn maintenance. The observed distribution of rills, wash marks, and leaf-debris dams demonstrates that only a small portion of a hillslope carries overland flow and is thus susceptible to sheetwash erosion. The susceptible sites are in drainage depressions where soils are shallow, and this suggests that flow is generated primarily by saturation of the entire soil profile during snowmelt and intense rainstonns. Published measurements of infiltration capacities for the soils also show that rain intensities alone are too low to generate overland flow on uncompacted, unsaturated hillslopes. Rates of sheet erosion and rilling were then estimated by applying the Universal Soil Loss Equa tion to road surfaces, construction sites, and the susceptible portions of hillslopes. Rilling rates were also estimated by measuring the length, cross-sectional area, and frequency of the rills that re-fonn each year during the runoff season. The rate calculated using the USLE is within a factor of two of that estimated from the rill measurements. Chapter 5: Applications 129

Dry ravel and sheet erosion also produce sediment from roadcuts. Measurements of datable root exposures on roadcuts were used to calculate retreat rates. Calculated rates were then applied to the area of roadcut in the watershed, which was evaluated by estimating roadcut heights at intervals along roads. Measurements on similar roads indicate that about 90% of this sediment accumulates at the base of the roadcut and is spread onto the road surface or adjacent hillsides during ditch main tenance (Reid 1981), and this value is supported by the extent of roadside deposits. Remobilization of these deposits would be accounted for by the s)1eetwash calcula tions. Thus, only a negligible 10% of the measured loss from roadcuts is included in the sediment budget as the estimated sediment production from roadcut erosion. The rootwad volumes and ages of treethrows were measured along streams and hillslope transects, and rates of sediment production to channels by treethrow were calculated from the observed frequency of channel-side treethrows. Because direct evaluation of bank erosion is difficult, we evaluated rates of soil transport on hillslopes to allow calculation of the rate of sediment supply to the banks. Transport processes include hillslope treethrow, animal burrowing, and soil creep; sheetwash also transports sediment, but most of this is contributed directly to channels. We measured the frequency of treethrows and rodent burrows along hillslope transects and added the transport rate resulting from these values to the average of published rates of soil creep for similar areas (Saunders and Young 1983). To estimate an average bank erosion rate, we then subtracted the measured rate of treethrow into the streams from the total transport rate on hillslopes. In several of the watersheds, recent meadow entrenchment also contributes sediment. Although incision of these valley-fill deposits represents remobilization of alluvial deposits, the conditions under which they accumulated no longer pertain. There thus is a net loss of sediment from alluvial storage, and this source must be considered in calculations of sediment production. The incision rate was estimated first by dividing the volume of the incised channel by the age of the incision (l00 years, estimated on the basis of land-use history), and then by applying the rate of bank retreat measured on roadcuts to the similar materials of the incised channels. Results of these two approaches agree to within a factor of two. A 17-year record of sedimentation exists for five weir ponds in two of the watersheds. Trap efficiency for the ponds was estimated from measured efficiencies of similar weirs on similar bedrock in Idaho (Megahan 1975). Our sediment pro duction estimates agree to within 10% with the measured average of 15 t/km2-yr. . The measured values are also within 10% of the average of 72 sediment yields measured in small basins with similar bedrock and climate throughout the western United States (Dendy and Champion 1978, Larson and Sidle 1980). Next, we needed to predict the effects of the planned hydro developments. Field observations of analogous developed sites showed that the only sediment source likely to be affected is sheetwash erosion on disturbed sites. The magnitude of this 130 Chapter 5: Applications change was estimated by applying the USLE to the planned construction sites and roads using parameters measured at analogous sites. Results suggest that distur bance currently accounts for 10% to 50% of the sediment inpurto channels in the six watersheds, and that disturbance-related rates would increase by an order of magnitude during the three years of construction. Most of the increase would be from road-surface erosion. The proposed projects would alter downstream hydrographs, and this would change the patterns of sediment transport through the streams. To evaluate ·the significance of these changes for fish habitat, we needed to characterize channel sediment transport under current and proposed conditions. We used published soil descriptions and our measurements of soil texture to compute the grain-size distri bution of sediment introduced into the streams (figure 12), and we evaluated the competence and capacity of channels to transport sediment using field observations of channel conditions. Small bars and patches of bed material of up to 256 mm in diameter lay in positions indicating recent channel transport. There was no obvious downstream trend in grain size or sorting, and no evidence of long-tern1 accumula tion, except on small fans where very steep tributaries entered the main river canyon. All particle sizes supplied to the streams were frequently mobile in the study watersheds. Sand was deposited in pools by the recession limbs of spring and summer flows, but was flushed downstream to become the suspended load of a larger river during high-flow peaks. Diversion of flow from floods capable of flushing sand-sized and coarser parti cles downstream would reduce the discharge by only a small fraction. For example, the present 2-year flood would become the 3-year flood after diversion, and the 25- year flood would become the 30-year flood. Thus, although diversion would de crease mean maximum monthly flows to 55-65% of present values, the bed material downstream of impoundments would continue to be flushed with only slightly longer intervals between transport events. However, pools and gravel interstices would probably become more pervasively filled with silt and sand between these flushing flows because of the increased contribution of fine sediments during project construction. The diversion impoundments would regularly fill withsedi ment and would need to be emptied. This could best be accomplished by equipping the impoundments with sluice gates that could be left open during seasons when mean monthly flows are too low to generate power but when the probability of flushing flows is relatively high.

5.2 Shinyanga Region, Tanzania An international organization planned to invest in development projects in north central Tanzania, but they needed to know what kinds of programs and which areas would provide the greatest increases in agricultural production. Soil erosion was Chapter 5: Applications 131 recognized as a major limit on agriculture, and the causes, effects, and distribution of the erosion had to be determined before potential control measures could be evaluated. The geomorphological phase of the study thus had the goals of deter mining: 1. The active erosion processes 2. Their distribution throughout the 21,000 km2 area 3. Their relative importance to sustainability of crop and livestock production 4. The influence of land use on erosion rates Because of the types of information required, we decided to adopt a sediment budgeting approach. The area was divided into subareas using wet- and dry-season Landsat imagery, and typical sites were visited within each subarea. The primary erosion processes were found to be sheetwash, gullying, and wind transport, and evidence for acceler ated erosion was found to increase with aridity. Within each subarea, apparent erosion intensity of all types increased with proximity to a village, and the onset of gullying was reported by residents to occur within about three years after a village was founded. Gullying was evaluated by measuring gully cross sections and lengths and interviewing local farmers to estimate rates of gully extension. The combined effects of sheetwash and wind erosion were evident from erosion mounds around datable acacia shrubs on rangelands and from the depth of erosional lag deposits on fallow fields of various ages. Wind and sheetwash erosion equations were also applied to each of the subareas using published climatic, topographic, and soils information, and field measurements of the percentage of ground covered by plants and coarse sediment. Soil-loss values predicted by the equations for cultivated areas were checked by comparing them to measurements from experimental plots in similar areas. Basic land-use information did not exist, so rangeland stocking densities were measured using systematic counting techniques during a low-level aerial survey, and the extent and distribution of various cultivation methods were determined by measuring their areas on periodic vertical photographs made during the survey (Norton-Griffiths 1983). This information was then used to distribute the loss rates associated with each land-use practice through each subarea. The study identified recent resettlement into villages as a major reason for the present episode of erosion; land use is now concentrated around centralized villages rather than being dispersed over widely separated farms. Because of its wide distribution, sheetwash erosion on rangelands is the primary source of sedimenta tion in reservoirs, but it will not thin the soil profile to an unworkable extent for many decades. Of greater importance are changes taking place on cultivated land. Farming techniques introduced over the past century generate sheetwash erosion at rates that will reduce the depth of arable upland soils from 35 cm to less than 15 cm 132 Chapter 5: Applications within 60 years. Calculations using soil-loss equations suggest that if the deep furrows and multi-cropping of traditional farming systems were restored, the life span of the A-horizon would be increased by a factor ofthree. Gullying is common in the area, but it is most significant as an agent of hydro logic change: the incised channels remove runoff more rapidly, so soil moisture receives less recharge and swale vegetation changes in response. Halting the growth of gullies would require major decreases in grazing intensity and changes in herding practices to increase ground cover density. These changes are not feasible under the present social and economic conditions in the area.

5.3 Snoqualmie River basin, Washington The 212-km2 South Fork Snoqualmie River basin in the Washington Cascade Range is much smaller than Shinyanga Region, but it is large compared to most management planning areas. The problem in this case was to estimate the probable sedimentation rate in proposed impoundments near the mouth of the basin. Not only did present sediment inputs need to be evaluated, but future conditions needed to be predicted so that the effects of continued development of the watershed's timber, recreation, and water resources could be anticipated. Preliminary fieldwork and examination of aerial photographs showed the major sediment sources to be landslides and roads in areas managed for timber production. The number and area of landslides that occurred over a 17-year period under each type of land use were measured using sequential aerial photographs, and field measurements provided a relationship between landslide areas and volumes. Soil samples were collected to determine the grain-size distribution of landslide debris, and from these data we computed the average annual landslide contribution of each grain size. Mean annual sediment yields and the grain size distribution of particles eroded from forest roads were estimated using results of field measurements made along roads in the Olympic Mountains (Reid and Dunne 1984). The estimated total influx was then partitioned into bedload and suspended load on the basis of grain size. Because ~and is present in soils of the area, its almost complete absence in point bars along the South Fork Snoqualmie indicates that sediment of sand size and smaller is flushed from the basin as suspended load. The estimated yields' of bedload and suspended load were checked against independent evidence. Nelson (1971) calculated annual suspended sediment yield from the South Fork Snoqualmie basin on the basis of two years of stream sampling and flow measurements, and estimated that bedload contributes 5% of the total load. The average for the two years was 60,000 t/yr of suspended load and 3,000 tlyr of bedload. Our values from the sediment budget were 42,000 and 2,600 t/yr, respectively. . We made another independent estimate of bedload when our measurements of bed-material texture indicated that all gravel bedload was accumulating on bars in a Chapter 5: Applications 133 low-gradient reach near the basin outlet, and only sand was leaving the basin. This depositional reach is part of a large, low-gradient fan accumulating at the Cascade front, and channel-bed sediments downstream of the fan consist only of sands. The areas of erosion and accretion of bars in the aggrading reach were measured on aerial photographs taken 14 years apart at similar low flows (Figure 2 in Dunne 1988). Changes in bar area were converted to the weight of deposited sediment by surveying typical bar cross sections with a hand level and tape and using a typical bulk density for poorly-sorted gravels. Bedload yield computed 'in this manner was 2,400 t/yr. Although the three independent methods used to estimate sediment yield each are susceptible to their own uncertainties, the agreement between the values lends credence to the result and supports the estimates of process rates used to construct the sediment budget. These values were then used to estimate sedimentation rates for the proposed impoundments, and they provided the baseline against which to compare the impacts of dam construction and maintenance. These latter calculations required construction of additional sediment budgets for construction and post construction phases. Because sediment production rates were calculated as a func tion of land-use practice, future rates could be estimated from projections of land use change in the basin.

5.4 Olympic Peninsula rivers, Washington Gravel extraction from river channels can alter channel morphology and down stream sediment supply, and these changes can degrade in-stream and riparian habitat, damage engineering structures, cause downstream scour, and alter patterns of flooding. To design sustainable, low-impact gravel mining plans, it would be useful to be able to estimate how much sediment can be extracted from a channel without exceeding the river's ability to replenish bars near the harvesting site. Elsewhere, aggrading channels may need to be dredged to maintain navigability and flood conveyance capacity. In this case, too, estimates of the sediment supply and the downstream morphological consequences of dredging would be useful. Both of these applications require that a sediment budget be constructed for a channel reach. Inputs from upstream and from within the reach would be evaluated, along with the discharge of sediment downstream and the artificial rates of extrac tion. The problem is simplified by the need to consider only bed material unless the planned excavation is deep enough to trap washload, but this possibility could be evaluated by methods described in a previous section. If there are no tributaries or significant bank erosion in the reach, the problem is further simplified: the amount of sediment that can be extracted annually without causing morphological change is simply the difference between the annual influx and discharge of bed material. Many reaches that are attractive for gravel harvesting, fish habitat, and recreation are zones of bed-material accumulation in which there is a strong decrease in 134 Chapter 5: Applications

sediment transport through the reach (Dunne 1988, p. 429), and the maximum equilibrium extraction rate may be nearly equal to the influx rate. Where there is appreciable gravel discharge from the reach, however, an extraction rate equal to the influx rate leaves the channel downstream starved of sediment, and scouring can ensue. It is commonly assumed that evaluations of sediment transport-must necessarily be expensive and time"consuming, so there has been little, if any, rigorous evalua" tion of the effects of sediment extraction from channels. Of course, it is possible to design a lengthy and expensive evaluation scheme by coupling an appropriate bed material transport equation to a steady-state flow model and then using repeated surveys of the reach to verify and calibrate the flow and transport equations. This is the basis for the HEC-6 model developed by the U.S. Army Corps of Engineers and used widely by the engineering community. Such a strategy is viable for continued management of high-value resources along critical reaches of large rivers, but it is rarely practical for routine, preliminary evaluation of smaller systems. Instead, in most cases the channel-reach sediment budget can be defined with sufficient precision by rapid and inexpensive methods using basic principles of fluvial geo morphology, existing measurements, and judicious collection of additional field data. Collins and Dunne (1989) evaluated the components of a long-term sediment budget for three gravel-bedded rivers that drain the southern Olympic Mountains in Washington. They first estimated the average annual discharge of bed material and its inter-annual variability by three independent methods: (1) using the Meyer-Peter and Muller (described by Graf 1971) and Parker et a!. (1982) bedload transport equations, being careful to apply them only to reaches that lie within the range of particle size and shear stress for which they were calibrated; (2) estimating bedload transport as a percentage of suspended load measured previously by the U.S. Geological Survey; and (3) measuring rates of bed-material accretion at bars on six sets of aerial photographs taken over a 44-year period. The results for one river are summarized in Figure 25, which reflects a relatively large influx of sediment from a tributary, and a sharp downstream reduction of bedload transport between river kilometers 3 and 21 followed by a more gradual downstream decrease. The sharp reduction could reflect the storage of approximately 4000 m 3/yr on the valley floor or the breakdown of sediment to suspendible sizes. Tumbling mill experiments on bed material and colluvium provided strong circumstantial evidence that attrition is responsible for a significant proportion of the change, and that the 4000 m 3/yr value is thus a maximum bound for the rate of harvestable accumulation. Records in the office of the responsible regulatory agency gave minimum values of gravel mining rates for periods of 25 to 40 years, together with their temporal variability. Extrac tion rates were at least an order of magnitude greater than the maximum potential rates of gravel accumulation on the valley floor. Chapter 6: Conclusions 135

C")--- E 7000 t 0 6000 a.. ·r--.... C/) c: / \ ....co 5000 { ...... "'0 I \ co I \ 0 4000 "'0 I \ Q) .0 3000 \ co \ :::l • \ c: 2000 • c: • \---1 co • Stevens Cr. Q) 0> 1000 tributary I co '- Q) > 0 <{ 0 5 10 15 20 25 30 35 40 45 Kilometers upstream of river mouth

Figure 25: Variation along the Humptulips River, Washington, of average annual rates of bedload transport estimated by two equations, and of bed-material accretion rates on gravel bars. Inter annual variations of bar accretion rates for the Humptulips, as well as estimates of transport and accumulation rates for neighboring rivers are given by Collins and Dunne (1989).

The imbalance between rates of sediment accumulation and off-take could have been accommodated by an increase of channel width and depth. The aerial photo records, engineering surveys at bridge crossings, and current-meter records show a clear spatial and temporal association between channel-bed lowering and intense gravel extraction, with an average channel incision rate of 3 cm/yr. The channel surveys were used to obtain approximate volumes of storage change in the exploited reaches, and computed volumes agree roughly with the difference between bed material influx and combined bed-material discharge and attrition. The details of this analysis are described by Collins and Dunne (1989). Collins and Dunne (1990) provide a broader analysis of other sediment-budget imbalances in valley reaches used for gravel extraction.

6 CONCLUSIONS

A document of this size cannot pretend to be a complete guide to sediment budget construction, but it can provide an overview of strategies and tools that are useful for understanding sediment production and transport in watersheds. The validity of a sediment budget then depends on how wisely these methods are employed. 136 Chapter 6: Conclusions

Effective construction and interpretation of sediment budgets requires a sound understanding of erosion and sedimentation processes, experience in field mapping and in the measurement and analysis techniques to be used, and, above all, good professional judgement. "Cookbook"-type instructional manuals for sediment budgeting do not exist and would not be appropriate. Each area presents its own difficulties and opportunities, so analysts must have a strong enough background in geomorphology and hydrology to take advantage of the peculiarities of the area to be evaluated, and they must be creative and open-minded in their approach. Sedi ment budgeting is not an approach that can be reduced to a set of instructions that can be carried out by those without training in geomorphology and hydrology. There have not yet been enough sediment budget studies to allow statistical evaluation of the accuracy and reproducibility of the general approach. However, two indications suggest that the method's components are sufficiently reproducible for many purposes. First, we have found that when several trained geomorpholo gists are asked to evaluate a process rate, results agree relatively closely, and certainly to well within an order of magnitude. Because many sediment budget applications require only approximate estimates, this level of accuracy is expected to be adequate. Second, sediment yields predicted by budgets often can be compared to meas ured values for similar areas or to values calculated using other approaches. In each case, results have agreed to within a factor of 2, and usually to within 30% (table 17). Often, however, the quality of available monitoring data is suspect or the monitoring record is too short to reflect a long-term average. If a sediment budget is constructed to include input from large events, while the monitoring record covers a period devoid of such events, then the constructed budget may give a better estimate of average sediment yields than the measurements. Differences between calculated, measured, and expected values should be viewed as an indication that further work must be done to resolve the differences, but there is rarely an a priori basis for assuming that a particular monitored value is the one that is correct. Construction of sediment budgets is more difficult in some areas than others. At sites where sediment input is dominated by large, infrequent events, rates must be evaluated using as long a period of record as possible. In such cases, land use may cause small changes in process frequencies which can strongly affect long-term sediment yields, but which may not be observable over the time frame available for analysis. For example, if clearcutting reduces the storm size necessary to trigger massive landslides from a 100-year to a 50-year recurrence interval, then long-term sediment yield may be effectively doubled. However, there may not be direct evidence of the change until a 50-year storm occurs. Sediment budgeting may not be able to quantify such changes, but it provides one of the few approaches capable of revealing their potential importance. Chapter 6: Conclusions 137

Table 17: Comparison ofmeasured sediment yields with those calculated using sediment budgets Sediment yield Monitoring (t_km-2_y(l) duration Site measured calculated (yr) Reference Teakettle Creek, CA 15 16 17 this paper Simas Valley, CA 242' 160 6 Reid 1989 Snoqualmie River, WA 283" 198" 2 Dunne 1988; this paper

. * measurements from similar basin 7 km away * * only suspended sediment component considered

The most difficult aspects of a sediment budget to quantify are those involving transport and storage of sediment in channels, and future research needs to focus on these components. Meanwhile, in areas where these components are particularly important, sediment budgets can often reveal the process interactions that control channel response, the types of changes a channel may undergo, and the likely location of those changes, even though the rates of change cannot be quantified. Applications for sediment budgeting range from analysis of on-going environ mental changes to prediction of changes likely to result from proposed management plans_ Information can be used to design projects that minimize environmental change, to plan the most effective restoration or soil conservation strategies in areas already undergoing degradation, and to evaluate future impacts of new develop ment. Sediment budget information can also be useful as a component of larger research projects. We have used approximate sediment budgets in an evaluation of sediment transport through gully systems (Reid 1989) and to establish background erosion rates for assessing the relative importance of road-related erosion in logged areas (Reid et al. 1981). Long-term monitoring of process rates was not possible in any of the cases described here. Instead, we used techniques that can be applied during reconnais sance-level fieldwork. We often used opportunistic field evidence to measure process rates, and we have found that such evidence is remarkably plentiful if one has clearly defined the quantities that need to be determined. In other cases, a basic understanding of the physical and biological controls on process rates allowed either the application of standard procedures to compute process rates or the use of published rate measurements to estimate rates in the project areas. The methods we describe have been applied successfully from the semi-arid rangelands of Tanzania to the partially logged rainforests of western Washington, and the approach has efficiently provided information critical for management dec1sions at each site. 138 Appendices

APPENDIX 1. GLOSSARY active sediment Channel sediment that is transported at least once every several years active transport Portion of channel bed that regularly transports sediment zone aggradation Accumulation of sediment, usually implying an increase in deposit thickness alluvial fan Fan-shaped accumulation of sediment deposited from flowing water alluvium Sediment deposited by flowing water anastomosing Channel pattern with mUltiple flow threads armor layer 1) Large sediment grains left on a hillslope surface because they can not be carried off by the active transport processes; or 2) a layer of coarse sediment on the surface of a channel bed that can not be transported by the existing flow regime (compare "pavement") bathymetry Subaquaeoustopography bed material Sediment present in channel-bed deposits bed-material Sediment of sizes present in channel-bed deposits, but which is suspended load transported in suspension bedload Sediment carried along a channel bed by sliding, rolling, or bouncing braided Channel pattern with multiple flow threads bulk density Mass of material per unit volume capacity In sediment transport, the potential maximum transport rate for a given flow catchment The area contained within the drainage divide above a speci fied point along a stream (equivalent to "watershed" or "drainage basin") channel geometry A description of the cross-sectional form of a channel; meas urements usually include width, depth, gradient, cross-sec tional area, and perimeter

channel order In the Strahler system, a headwater channel is of order 1, and when two channels of like order join, they create a channel of one higher order (e.g. 1+1=2, 1+2=2, 2+2=3, etc.) Appendices 139 chemical The removal of dissolved material from an area denudation chronic erosion Erosion processes that are frequently active at the same sites, processes and are usually widespread across a landscape colluvium Soil and rock debris on a hillslope that has been transported from its original location competence In sediment transport, the largest size of particle that a given flow can transport critical shear Minimum shear stress necessary to move a given particle stress cumulative Combined environmental effects of multiple land-use activities watershed effect in a watershed, where effects involve watershed processes debris avalanche Rapidly moving shallow landslide that usually displaces colluvium, saprolite, and only minimal bedrock debris fan Fan-shaped accumulation of eroded sediment, rock debris, and organic material debris flow Rapidly moving flow of sediment, rock debris, and organic material delivery ratio Ratio between the amount of sediment leaving a transport link (such as a channel reach, channel system, or length of hill slope) and the amount originally contributed to the link denudation Loss of primary rock material from an area; often the denu dation rate is expressed as a rate of ground-surface lowering discrete erosion Erosion processes that usually occur once or infrequently at processes any particular site; events can thus be enumerated discriminant Statistical technique used to define factors that best permit analysis discrimination between groups dissolution Process that dissolves material drainage basin The area contained within the drainage divide above a speci fied point along a stream (equivalent to "watershed" or "catchment") dry ravel Sloughing of sediment due to loss of cohesion in surface materials earthflow Slow flow of moist, usually clay-rich colluvium equal mobility In sediment transport, the concept that 1) most grain sizes in a channel bed are mobilized by the same critical shear stress, 140 Appendices

or 2) the transport rates of most grain sizes in a channel reach are approximately equal erOSIOn Removal of sediment or rock from a point in a landscape evapo Release of water vapor to the atmosphere by the combination transpiration of direct evaporation and transpiration by plants floodplain Low-gradient surfaces along a channel that are occasionally inundated by floods; often implies that the surfaces are formed by accumulation of channel deposits geographic Computer software that organizes and allows manipUlation of information spatially distributed data (GIS; note that "GIS system" is system redundant) geomorphology The study of physical landscapes and the processes that mold them gleying The discoloration of clays in a soil by reduction under moist conditions; grey clay horizons thus often imply waterlogged soil conditions graticule loupe A magnifying lens with an inscribed measuring scale gully An incised channel larger than a rill; the size distinction is not formal, and has been variously defined as "too large to be obliterated by plowing" and "wider than 12 inches" headcut A vertical break in a channel profile composed of erodible material. The term usually implies that the feature is eroding headward, though a headcut at the head of a channel can be essentially stationary. hydraulic radius A measure of the average distance of flow from the bed or banks of a channel; computed by dividing the cross-sectional area of a channel by its wetted perimeter. It is approximately equal to the average depth in most channels. hydro graph The distribution of runoff through time at a point imbrication In sedimentology, the leaning of sediment clasts against one another such that they are tilted in the direction of flow interception In hydrology, the accumulation of precipitation on vegetation and other above-ground surfaces and its evaporation during and after the storm isohyet A line on a map, along which all points receive the same amount of precipitation Appendices 141 karst Landscape influenced by dissolution of limestones; usually characterized by caves and sinkholes lithology The type of rock loess Silts accumulated by wind deposition McNeil sampler Cylindrical tube used to collect samples of channel-bed sediments median grain A grain size that is coarser than half of the sediment volume in diameter a sample and finer than the other half overland flow Unchannelled flow of water over the ground surface pavement In sedimentology, a surficial deposit of sediment coarser than the underlying sediments, but that can be transported regu larly by the existing flow regime (compare "armor layer") photogrammetry Measurements made on photographs; usually implies that aerial photographs are used piping A term often used for the erosion of tunnels in soils by subsur face flow. ''Tunnel erosion" is a better term for this process because, in the engineering literature, "piping'" refers to seepage erosion. plungepool Pool carved by water falling over an obstruction or headcut

~oint count Method of estimating the proportion of an Cjrea covered by particular features on the basis of a statistical sampling of observed points. Thus, if 20% of the points randomly sam pled in a clearfelled tract are found to be compacted, 20% of the tract is assumed to be compacted. quickflow Storm runoff that reaches a channel within hours of rainfall rainsplash Displacement of sediment by bombardment of raindrops recurrence interval Average length of time between successive events

regolith Loose material on the earth's surface, including saprolite, colluvium, and alluvium. Usually implies that the materials are currently subaerial, though they may have been deposited subaqueously. relief The maximum elevation difference within a given area residual soil 1) Soil formed by weathering of bedrock at a site; or 2) more complete soil profile preserved in an area where most pro files have been depleted by erosion 142 Appendices rill An incised channel smaller than a gully; the size distinction is not formal, and has been variously defined as "small enough to plow over" and "narrower than 12 inches" saprolite In-situ strongly weathered bedrock that has decomposed to an earthy consistency sediment budget An accounting of the sources and disposition of sediment as it travels from its point. of origin to its eventual exit from a drainage basin sediment delivery Contribution of transported sediment to a particular location or part of a landscape sediment Delivery of sediment from hiJIslopes to channels production sediment storage Deposition of sediment after a period of transport sediment yield Amount of sediment passing a particular point in a watershed per unit of time sedimentation Deposition of sediment sequential aerial Aerial photographs taken of an area at different times. These photographs usually provide the most effective tool for evaluating major landscape changes through time. settling velocity Velocity at which a particle falls through a fluid shear stress The force per unit area exerted in the direction of flow by flow parallel to a surface sheet erosion Erosion of the ground surface by unconcentrated (i.e. not in rills) overland flow sheetwash Sheet erosion slump Relatively deep landslide occuring along a curved slip surface, such that the surface of the displaced block is rotated slightly backward from its original orientation soil 1) In pedology, slope materials altered by physical and chemi cal weathering due to their presence at or near the ground surface; often implies that chemically or texturally defined horizons are at least incipiently developed, or that plants have modified the material present. 2) In engineering, any subaerial loose material; equivalent to regolith soil creep Slow down-slope transport of soil due primarily to the influ ence of gravity on many small disturbances, but also includ ing slow plastic flow Appendices 143 solifluction Downslope flowage of saturated soil, usually ill permafrost ten-ain. Where the motion is induced by thawing of soil ice, the term "gelifluction" is often used. stratification 1) Layering in sedimentary rock. 2) Division of a population or area into subsets of uniform character for sampling stream power Available power in a stream per unit length of channel bed, calculated as the product of discharge and gradient suspended load Sediment can-ied within the water column of a stream talus Loose rock debris accumulated by falling, rolling, and sliding, usually at the foot of a rock face ten-ace A low-gradient surface formed by fluvial aggradation or erosion when the stream flowed at a higher elevation in the landscape. The term usually implies that the surface is rarely, if ever, inundated by floods in the cun-ent hydrologic regime. thalweg A line along the bed of a channel that connects the deepest points of each cross section through-put load A fraction of the bed-material load present in some channels that moves across the surface of the channel-bed pavement gravels during flows too small to mobilize the pavement gravels, and which has grain sizes smaller than those of the pavement. total load 1) In sediment transport equations, the sediment load in a channel exclusive of washload. 2) Otherwise, all sediment load in a channel, including bed-material load and washload. transition Table that displays the likelihood of transfer from one category probability (e.g. those listed in rows) to each of several others (e.g. those matrix listed in columns) trap efficiency Proportion of sediment influx to a water body that is deposited within the body treethrow erosion Displacement of soil held by the roots of a toppling tree; uprooting tunnel erosion Erosion of pipes in soils by subsurface flow; piping erosion uprooting Displacement of soil held by the roots of a toppling tree; treethrow erosion USLE Universal Soil Loss Equation: a method for calculating rates of sheet and rill erosion on agricultural and range lands 144 Appendices washload Sediment carried in suspension by flow, and which is of sizes not represented in the bed material watershed The area contained within the drainage divide above a speci fied point along a stream (same as catchment, drainage basin) WEPP Water Erosion Prediction Procedure: a method for calculating rates of sheet and rill erosion on agricultural, range, and forest lands, and on r\)ad surfaces.

APPENDIX 2. ADDITIONAL READING AND USEFUL COMPENDIA

Background information Birkeland, P.W. 1984. Soils and Geomorphology. Oxford University Press. 372 pp. Good general overview ofsoil classification, formation processes, evolution, and relation to geomorphological processes. Throughout, applications of soil science to the solution ofgeomorphological problems are stressed. Carson, M.A., and G.A. Griffiths. 1987. Bedload transport in gravel channels. Journal of Hydrology (New Zealand) 26(1):1-15l. Excellent discussion of how gravel-bedded channels work, with particular relevance for braided channels. Topics include: initiation of motion, measurement qnd prediction of bedload transport, and interactions between channel form and gravel transport. Carson, M.A., and M.J. Kirkby. 1972. Hillslope Form and Process. Cambridge University Press. 475 pp. Classic textbook describing materials and forces on hills lopes, and the processes that result from their interaction. Includes discussions of the mechanisms, analyti cal methods for, and rates of landsliding, surface water erosion, subsurface water erosion, and soil creep. Also discusses the hills lope forms resulting from each process in different climatic zones, and applies hills lope simulation models to explore their effects. Dunne, T., and L.B. Leopold. 1978. Water in Environmental Planning. W.H. Freeman and Co. 818 pp. The first section of this textbook works through each step of the hydrologic cycle in chapters on precipitation, interception, evapotranspiration, soil and groundwa ter, runoff, flooding, water supply, and snow hydrology. This is followed by sections on geomorphology (including hills lope and channel processes) and water quality. In each case, practical field techniques and office procedures for evaluation and analysis are described, fInd worked examples are given. Appendices 145

Goudie, A. (ed.). 1981. Geomorphological Techniques. George Allen and Unwin. 395 pp. This book is a collection of descriptions of standard techniques used in measur ing and describing landforms, hills lope and channel materials, cmd erosion and sediment transport processes; it also includes a section on dating techniques. Most methods that are included for process rate assessments involve monitoring rates through time, and so are not particularly useful for reconnaissance-level work.

Graf, W.H. 1971. Hydraulics of Sediment Transport. McGraw-Hill Book Co. 513 pp. A comprehensive, comprehensible review of transport mechanics and many of the predictive and measurement methods still in use. In particular, the book in cludes useful discussions of various sediment transport equations. Gregory, K.J., and D.E. Walling. 1973. Drainage Basin Form and Process. Halsted Press. 458 pp. Good overview of geomorphology and hydrology relevant to a watershed scale. Includes discussions of drainage basin characteristics, hydrological measurement techniques, monitoring methods for erosion and sediment transport, hillslope and channel processes, and changes through time. The book provides abundant refer ences for more detailed coverage of each topic, and compiles many measurements from diverse sources into comparative tables. Herschy, R.W. 1985. Streamflow Measurement. Elsevier Applied Science Publish ers, London. 553 pp. Provides detailed iriformation on many different methods for calculating stream discharge: velocity-area, stage, weirs, dilution, and others. Techniques and equip ment for measuring current velocity are also described. Contains worked examples. Richards, K. 1982. Rivers: Form and Process in Alluvial Channels. Methuen, London. 358 pp. Describes sediment transport mechanisms, measurement, and analysis proce dures; and discusses the processes controlling channel form, gradient, and adjust ments to change. Good overview offluvial geomorphology, and an excellent guide to other literature on the topic. Ritter, D.F. 1986. Process Geomorphology (Second edition). William C. Brown, Dubuque, Iowa. 579 pp. A good general textbook on geomorphology. It differs from many older geomor phology texts in stressing processes and mechanisms responsible for shaping landscapes, rather than merely describing the landforms present. Selby, M.J. 1982. Hillslope Materials and Processes. Oxford University Press. 264 pp. This book contains more iriformation on hills lope materials and weathering than 146 Appendices the text by Carson and Kirkby. The book carries a particular emphasis on rock slope processes and landslides, and provides detailed discussions of mechanisms and analytic methods for these processes. Water erosion processes on hills lopes are more briefly covered, and do not include descriptions ofanalytical techniques.

Swanson, F.J.; R.J. Janda, T. Dunne, and D.N. Swanston. 1982. Sediment budgets and routing in forested drainage basins. U.S.D.A. Forest Service General Technical Report PNW-141. 165pp. A proceedings volume from an early conference on sediment budgeting that demonstrates a variety of approaches, components, and applications. The paper by Dietrich et al. provides a general discussion of the elements of a sediment budget and the concepts behind budget construction. Vanoni, V.A. (editor). 1975. Sedimentation Engineering. American Society of Civil Engineers, New York. 745 pp. This essentially is a manual of methods for addressing a variety of sediment related problems. It contains good background information on the causes and mechanisms of erosion, sediment transport and aggradation, and examines a variety of standard methods for evaluating each through predictive equations or monitoring. Included is a good description of the use, limitations, and basis for a variety of sediment transport equations. Finally, methods of controlling sediment and the legal implications ofsediment problems are discussed. Way, D.S. 1978. Terrain Analysis. Dowden, Hutchinson, and Ross; Stroudsburg, Pennsylvania. 438 pp. The book could be subtitled "Air-photo analysis for land-use planning". Geo logical, cultural, and geomorphological air-photo interpretation are covered, and abundant illustrations and tables are provided to supplement the quite readable text. Wolf, P.R. 1974. Elements of Photo gramme try. McGraw-Hill, Inc.; New York. 562 pp. This textbook provides excellent backgroundfor using aerial photographs, and is particularly useful if you need to contract a new flight or are planning to make detailed measurements from photographs.

Manuals and descriptions of particular methods Barnes, H.H. 1967. Roughness characteristics of natural channels. U.S. Geological Survey Water Supply Paper 1849. 213 pp. Contains photographs of river channels with measured Mannings' n values, allowing n to be estimated for similar channels. Appendices 147

Benson, M.A., and T. Dalrymple. 1967. General field and office procedures for indirect discharge measurements. Techniques of Water-Resources Investigations of the V.S. Geological Survey. Book 3, Chapter AI. Pp. 1-30. Describes USGS procedures for making channel measurements, and describes methods of estimating peak discharge from evidence remaining after floods.

Church, M.A; D.G. McLean, and J.F. Wolcott. 1987. River bed gravels: sampling and analysis. Pp. 43-88 in C.R. Thome, J.C. Bathurst, and R.D. Hey (eds.): Sedi ment Transport in Gravel-Bed Rivers. John Wiley and Sons, New York. 995 pp. Describes procedures for measuring grain-size distributions in gravel-bedded channels, with graphs that allow estimation of necessary sample sizes. Dunne, T. 1977. Evaluation of erosion conditions and trends. Pp. 53-83 in: S.H. Kunkle and J.L. Thames (editors), Guidelines for Watershed Management. FAO Conservation Guide 1, UN Food and Agriculture Organization, Rome. A guide to methods for measuring erosion rates, including reconnaissance-level techniques, monitoring methods, and in-stream measurements. MacDonald, L.H., AW. Smart, and R.C. Wissmar. 1991. Monitoring Guidelines to evaluate effects of forestry activities on streams in the Pacific Northwest and Alaska. U.S. Environmental Protection Agency, Region 10. EPA/910/9-91-00I. 166 pp. Description and discussion of methods available for monitoring variables associated with water quality, including water chemistry, water temperature, flow, sediment load, channel characteristics, riparian vegetation, and aquatic organisms. Wischmeier, W.H., and D.O. Smith. 1978. Predicting rainfall erosion losses-a guide to conservation planning. U.S.D.A. Agricultural Handbook 537.58 pp. The manual for application ofthe Universal Soil Loss Equation.

Data compendia Dendy, F.E., and W.A. Champion. 1978. Sediment deposition in U.S. reservoirs. Summary of data reported through 1975. V.S.D.A. Miscellaneous Publication 1362. 84 pp. Tables ofreservoir sedimentation data from the continental United States. Larson, K.R., and R.C. Sidle. 1980. Erosion and sedimentation data catalog of the Pacific Northwest. V.S.D.A Forest Service, Pacific Northwest Region; R6-WM- 050-1981; 64 pp. Description of studies that have provided information on logging-related sedi ment yields in the Pacific Northwest (including California). Includes results and site descriptions. 148 Appendices

NCAS1. 1985. Catalog of landslide inventories for the Northwest. National Council of the Paper Industry for Air and Stream Improvement; Technical Bulletin 456. 78 pp. Compendium of landslide surveys in the Pacific Northwest. This is a useful directory to the types and locations of studies, but includes some inaccuracies in their descriptions. Primary sources should be examined.

Saunders, 1., and A. Young. 1983. Rates of surface processes on slopes, slope retreat, and denudation. Earth Surface Processes and Landforms 8:473-501. Compilation of measured rates of soil creep, solifluction, surface wash, dissolu tion, rock weathering, slope retreat, cliff retreat, and sheetwash. References 149

REFERENCES

Abel, N. , and M. Stocking. 1987. A rapid method for assessing rates of soil erosion from rangeland: an example from Botswana. Journal of Range Management 40(5):460-466. Ackers, P., and W.R. White . 1973. Sediment transport: a new approach and analysis. Journal of the Hydraulics Division, American Society of Civil Engineers 99(HYII):2041-2060. Alonso, e.Y.; W.H. Neibling, and G.R. Foster. 1981. Estimating sediment transport capacity in watershed modeling. Transactions of the American Society of Agricultural Engineers 24(5): 1211-1220, 1226. Amin, M.I., and PJ. Murphy. 1981. Two bed-load formulas: an evaluation. Journal of the Hydrau lics Division, American Society of Civil Engineers 107(HY8):961-972. Andersen, D.e. 1987. Geomys bursarius burrowing patterns: influence of season and food patch structure. Ecology 68(5): 1306-1318. Anderson, E.W. 1974. Indicators of soil movement on range watersheds. Journal of Range Man agement 27(3):244-247. Andrews, E.D. 1979. Scour and fill in a stream channel, East Fork River, Western Wyoming. U.S. Geological Survey Professional Paper 1117.49 pp. Andrews, E.D. 1981. Measurement and computation of bed-material discharge in a shallow sand bed stream, Muddy Creek, Wyoming. Water Resources Research 17(1):131-141. Andrews, E.D. 1983. Entrainment of gravel from naturally sorted riverbed material. Geological Society of America Bulletin 94(10): 1225-1231. Argabright, M.S. 1991. Evolution in use and development of the wind erosion equation. Journal of Soil and Water Conservation 46(2): 104-106. Ashmore, P.E.; T.R. Yuzyk, and R. Herrington. 1989. Bed material sampling error in sand bed rivers. Water Resources Research 25(10):2195-2204. Ashworth, PJ., and R.1. Ferguson. 1989. Size-selective entrainment of bed load in gravel bed streams. Water Resources Research 25(4):627-634. Bagnold, R.A. 1956. The flow of cohesionless grains in fluids . Transactions of the Royal Society (London) 249:235-297. Bagnold, R.A. 1966. An approach to the sediment transport problem from general physics. U.S. Geological Survey Professional Paper 422-1. 37 pp. Bagnold, R.A. 1980. An empirical correlation of bedload transport rates in flumes and natural rivers. Proceedings of the Royal Society (London) 372:452-473. Barnes, H.H. 1967. Roughness characteristics of natural channels. U.S. Geological Survey Water Supply Paper 1849.213 pp. Bathurst, J.e.; W.H. Graf, and H.H. Cao. 1987. Bed load discharge equations for steep mountain rivers. Pp. 453-491 in C.R. Thome, J.e. Bathurst, and R.D. Hey (eds.): Sediment Transport in Gravel-Bed Rivers. John Wiley and Sons, New York. 995 pp. Bathurst, J.C.; GJ.L. Leeks, and M.D. Newson. 1986. Relationship between sediment supply and sediment transport for the Roaring River, Colorado, U.S.A. Pp. 105-117 in R.F. Hadley (ed.): Drainage Basin Sediment Delivery. International Association of Hydrological Sciences Publi cation 159. Benda, L., and T. Dunne. 1987. Sediment routing by debris flows. Pp. 213-223 in Beschta, R.L., T. Blinn, G.E. Grant, G.G. Ice, and F.J. Swanson (eds.): Erosion and Sedimentation in the Pacific Rim. International Association of Hydrological Sciences Publication 165 150 References

Benson, M.A., and T. Dalrymple. 1967. General field and office procedures for indirect discharge measurements. Techniques of Water-Resources Investigations of the U.S. Geological Survey. Book 3, Chapter A I. Pp. 1-30. Bergstrom, F.W. 1982. Episodic behavior in badlands: its effects on channel morphology and sediment yields. Pp. 59-66 in: Sediment Budgets and Routing in Forested Drainage Basins. U.S.D.A. Forest Service General Technical Report PNW-141. 165 pp. Bishop, A.A.; D.B. Simons, and E.V . Richardson. 1965. Total bed-material transport. Proceedings of the American Society of Civil Engineers 91(HY2). Blackburn, W.H. 1980. Universal Soil Loss Equation and rangelands. Pp. 588-597 in: Symposium on Watershed Management 1980. American Society of Civil Engineers New York. 597 pp. Blench, T. 1966. Mobile-bed Fluviology. Department of Technical Services, University of Alberta, Edmonton, Alberta. Blong, RJ.; O.P. Graham, and J.A . Veness. 1982. The role of sidewall processes in gully develop ment, some NSW examples. Earth Surface Processes and L;mdforms 7:381-385. Bodhaine, G.L. 1968. Measurement of peak discharge at culverts by indirect methods. Chapter A3, Book 3, of Techniques of Water Resources Investigations of the U.S. Geological Survey. 60 pp. Bondurant, D.e. 1958. The total sediment load of streams: a discussion. Proceedings of the Ameri can Society of Civil Engineers 84(HY6). Bondy, E.L.; L. Lyles, and W.A. Hayes. 1980. Computing soil erosion by periods using wind energy distribution. Journal of Soil and Water Conservation 35(4): 173-176. Booth, D.B. ; K. Bell, and K.X. Whipple. 1991. Sediment transport along the South Fork and mainstem of the Snoqualmie River. King County Surface Water Management Division, Basin Planning Program, Seattle, Washington. 43 pp. Borah, D.K. 1989. Scour depth prediction under armoring conditions. Journal of Hydraulic Engi neering 115(10):1421-1425. Bradley, w.e. 1970. Effect of weathering on abrasion of granitic gravel, Colorado River (Texas). Geological Society of America Bulletin 81 :61-80. Brown, e.B. 1950. Sediment transportation. Chapter 12 in H. Rouse (ed.): Engineering Hydraulics. John Wiley and Sons, New York. 1039 pp. Brune, G.M. 1953. Trap efficiency of reservoirs. American Geophysical Union Transactions 34(3):407-418. Bull, K.R.; J.R. Hall, and R.G.H. Bunce. 1988. The use of physical parameters for selecting representative samples of river catchments. Journal of Environmental Management 27(4):405- 420. Burroughs, E.R., Jr. 1984. Landslide hazard rating for portions of the Oregon Coast Range. Pp. 265- 274 in C.L. O'Loughlin and AJ. Pearce (eds.): Proceedings of a Symposium on Effects of For est Land Use on Erosion and Slope Stability. 7-1 I May 1984. East-West Center, University of Hawaii, Honolulu, Hawaii. 310 pp. Caine, N. 1980. The rainfall intensity-duration control of shallow landslides and debris flows. Geografiska Annaler 62A(I-2):23-27. Caine, N., and F.J. Swanson. 1989. Geomorphic coupling of hills lope and channel systems in two small mountain basins. Zeitschrift fijr Geomorphologie N.F. 33(2): 189-203. Campbell, R.B. , Jr. 1981. Field and laboratory methods for age determination of quaking aspen. U.S .D.A. Forest Service Research Note INT-314. 5 pp. Carey, W.P. 1985. Variability in measured bedload-transport rates. Water Resources Bulletin 21 (I ):39-48. References 151

Carling, P.A. 1987. Bed stability in gravel streams, with reference to stream regulation and. ecology. Pp. 321-347 in K. Richards (ed.): River Channel: Environment and Process. The Institute of British Geographers, Special Publication. Carling, P.A. 1988. The concept of dominant discharge applied to two gravel-bed streams in relation to channel stability thresholds. Earth Surface Processes and Landforms 13(4):355-367. Carling, P.A ., and N .A. Reader. 1982. Structure, composition and bulk properties of upland stream gravels. Earth Surface Processes and Landforms 7(4):349-366. Carrarra, P.E., and T.R. Carroll. 1979. The determination of erosion rates from exposed tree roots in the Piceance Basin, Colorado. Earth Surface Processes 4:307-318. Carson, M.A., and G. A. Griffiths. 1987. Bedload transport in gravel channels. Journal of Hydrology (New Zealand) 26:1-151. Carson, M.A., and G. A. Griffiths. 1989. Gravel transport in the braided Waimakariri River: mechanisms, measurements, and predictions. Journal of Hydrology 109(3/4):201-220. Chang, M. ; S.P. Watters, and A.K. Sayok. 1989. A comparison of methods of estimating mean watershed slope. Water Resources Bulletin 25(2):327-334. Chow, V.T. (ed.). 1964. Handbook of Applied Hydrology. McGraw-Hili, New York. Church, M.A. 1978. Palaeohydrological reconstruction from a Holocene valley fill. In A.D. Miall (ed.): Fluvial Sedimentology. Canadian Society of Petroleum Geologists Memoir 5. Pp.743- 772. Church, M.A.; D.G. McLean, and J.F. Wolcott. 1987. River bed gravels: sampling and analysis. Pp. 43-88 in C.R. Thome, J.c. Bathurst, and R.D. Hey (eds.): Sediment Transport in Gravel-Bed Rivers. John Wiley and Sons, New York. 995 pp. Church, M., and O. Slaymaker. 1989. Disequilibrium of Holocene sediment yield in glaciated British Columbia. Nature 337(6206):452-453. CIWD. 1977. Sediment data Canadian Rivers 1975 . Canada Inland Waters Directorate, Water Resources Branch, Department of Fisheries and Environment, Ottawa, Canada. 364 pp. Colby, B.R. 1964. Discharge of sands and mean velocity relationships in sand-bed streams. U.S. Geological Survey Professional Paper 462-A. Colby, B.R.; and C.H. Hembree. 1955. Computations of total sediment discharge, Niobrara River near Cody, Nebraska. U.S. Geological Survey Water Supply Paper 1357. Collins, B.D., and T. Dunne. 1989. Gravel transport, gravel harvesting, and channel-bed degradation in rivers draining the southern Olympic Mountains, Washington, U.S.A. Environmental Geol ogy and Water Science 13(3):213-224. Collins, B.D., and T. Dunne. 1990. Assessing the effects of gravel harvesting on sediment transport and channel morphology: a guide for planners. State of California Division of Mines and Geol ogy, Sacramento, CA. 26 pp. Cooke, R.U., and R.W. Reeves. 1976. Arroyos and Environmental Change in the American South West. Clarendon Press, Oxford. 213 pp. Corbett, E.S., and R.M. Rice. 1966. Soil slippage increased by brush conversion. U.S.D.A. Forest Service Research Note PSW-128. 8 pp. Costa, J.E. 1983. Paleohydraulic reconstruction of flash flood peaks from boulder deposits in the Colorado Front Range. Geological Society of America Bulletin 94:986-1004. Crouch, R.J. 1987. The relationship of gully sidewall shape to sediment production. Australian Journal of Soil Research 25 :531-539. Daniels, R.B.; J.W. Gilliam, O.K. Cassel, and L.A. Nelson. 1985. Soil erosion class and landscape position in the North Carolina Piedmont. Soil Science Society of America Journal 49:991-995. 152 References

Darrach, A.G.; N.M. Curtis, Jr., and W.J. Sauerwein. 1978. Estimating sheet-rill erosion and sediment yield on rural and forest highways. U.S.D.A. Soil Conservation Service West Techni cal Service Center, Portland, Oregon; Technical Note Woodland-12. 42 pp. Day, W.R. 1950. The soil conditions which determine windthrow in forests. Forestry 23:90-95. Dendy, F .E. 1974. Sediment trap efficiency of small reservoirs. Transactions of the American Society of Agricultural Engineers 17(5):898-901,908. Dendy, F.E. 1982. Distribution of sediment deposits in small reservoir. American Society of Agricultural Engineers Transactions 25( I): 100-1 04. Dendy, F.E., and W.A. Champion. 1978. Sed.iment deposition in U.S. reservoirs. Summary of data reported through 1975. U.S.D.A. Miscellaneous Publication 1362.84 pp. Dendy, F.E., and C.M. Cooper. 1984. Sediment trap efficiency of a small reservoir. Journal of Soil and Water Conservation 39(4):278-280. Denny, C.S., and J.C. Goodlett. 1956. Micro-relief resulting from fallen trees. Pp. 59-66 in U.S. Geological Survey Professional Paper 288. Dickinson, W.T.; R.P. Rudra, and G.1. Wall. 1989 .. Nomographs and software for field and bank erosion. Journal of Soil and Water Conservation 44(6):596-600. Dietrich, W.E. 1982. Settling velocity of natural particles. Water Resources Research 18: 1615-1626. Dietrich, W.E., and T. Dunne. 1978. Sediment budget for a small catchment in mountainous terrain. Zeitschrift flir Geomorphologie Neue Folge Supp1ementband 29: 191-206. Dietrich, W.E., T. Dunne, N. Humphrey, and L.M. Reid. 1982. Construction of sediment budgets for drainage basins. Pp. 5-23 in Sediment Budgets and Routing in Forested Drainage Basins. U.S.D.A. Forest Service General Technical Report PNW-141. Dietrich, W.E., 1.W. Kirchner, H. Ikeda, and F. Iseya. 1989. Sediment supply and the development of the coarse surface layer in gravel-bedded rivers. Nature 340(6230):215-217. Diplas, P. 1987. Bedload transport in gravel-bed streams. Journal of Hydraulic Engineering 113(3):277-291. Dissmeyer, G.E. 1976. Erosion and sediment from forest land uses, management practices, and disturbances in southeastern United States. Pp. 1-140 to 1-148 in: Proceedings of the Third Fed eral Inter-Agency Sedimentation Conference, March 22-25, 1976, Denver CO. Sedimentation Committee, Water Resources Council. Dissmeyer, G.E., and G.R. Foster. 1981. Estimating the cover-management factor (C) in the Universal Soil Loss Equation for forest conditions. Journal of Soil and Water Conservation 36(4):235-240. Duijsings, J.J.H.M. 1986. Seasonal variation in the sediment delivery ratio of a forested drainage basin in Luxembourg. Pp. 153-164 in R.F. Hadley.(ed.): Drainage Basin Sediment Delivery. International Association of Hydrological Sciences Publication 159. 487 pp. Duijsings, J.J.H.M. 1987. A sediment budget for a forested catchment in Luxembourg and its implications for channel development. Earth Surface Processes and Landforms 12(2): 173-184. Dunne, T. 1977. Evaluation of erosion conditions and trends. Pp. 53-83 in: S.H. Kunkle and J.L. Thames (editors), Guidelines for Watershed Management. UN Food and Agriculture Organiza tion, Rome. FAO Conservation Guide I. Dunne, T. 1988. Contributions of geomorphology to flood-control planning. Pp. 421-438 in: V.R. Baker, R.C. Kochel, and P.C. Patton (editors), Flood Geomorphology. Wiley & Sons, N.Y. Dunne, T., B. Aubry, and E.K. Wahome. 1981. Effect of woodfuel harvest on soil erosion in Kenya. Report to the Ministry of Energy, Nairobi, Kenya. References 153

Dunne, T. , W.E. Dietrich, and MJ. Brunengo. 1978. Recent and past erosion rates in semi-arid Kenya. Zeitschrift fur Geomorpho10gie Neue Folge Supplementband 29: 130-140. Dunne, T. , W.E. Dietrich, and M.J. Brunengo. 1979. Rapid evaluation of soil erosion and soil lifespan in the grazing lands of Kenya. International Association of Hydrological S"iences Pub lication 128. Pp. 421-428. Dunne, T. , and L.B. Leopold. 1978. Water in Environmental Planning. W.H. Freeman and Co., San Francisco. 818 pp. Dunne, T.; T.R. Moore, and C.H. Taylor. 1975. Recognition and prediction of runoff-producing zones in humid regions. Hydrological Sciences Bulletin 20(3):305-327. Dynesius, M., and B.G. Jonsson. 1991. Dating uprooted trees: comparison and application of eight methods in a boreal forest. Canadian Journal of Forest Research 21 :655-665. Ebisemiju, F.S . 1990. Sediment delivery ratio prediction equations for short catchment slopes in a humid tropical environment. Journal of Hydrology 114(1-2):191-208. Einstein, H.A. 1950. The bedload function for sediment transportation in open channels. U .S.D.A. Soil Conservation Service Technical Bulletin 1026. Einstein, H.A. , and N. Chien. 1955. Effects of heavy sediment concentration near the bed on velocity and sediment distribution. U.S. Army Corps of Engineers, Missouri River Division Sediment Series number 8, Omaha, Nebraska, August 1955. Emery, K.A. 1975. Identification of soil erosion from aerial photographs. Journal of the Soil Conservation Service of New South Wales 31 (3):219-223, 235-240. Evans, R.A., and R.M. Love. 1957. The step point method of sampling - a practical tool in range research. Journal of Range Management 10:208-212. Everitt, B.L. 1968. Use of cottonwood in an investigation of the recent history of a floodplain. American Journal of Science 266:417-439. Farmer, E.E. and J.E. Fletcher. 1977. Highway erosion control systems: an evaluation based on the Universal Soil Loss Equation. Pp. 12-21 in : Soil Erosion: Prediction and Control. U.S. Soil Conservation Service Special Publication 21. Fleming, R.W., and A.M. Johnson. 1975. Rates of seasonal creep of silty clay soil. Quarterly Journal of Engineering Geology 8: 1-29. Foggin, G.T., and R.M. Rice. 1979. Predicting slope stability from aerial photos. Journal of Forestry 77(3): 152-155. Froelich, H.A., and M.R. Pyles. 1987. Rates of landsliding as impacted by timber management activities in northwestern California: Discussion. Bulletin of the Association of Engineering Geologists 24(3):425-431. Galatowitsch, S.M. 1990. Using the original land survey notes to reconstruct presettlement land scapes in the American west. Great Basin Naturalist 50(2): 181-192. Germanoski, D., and D.F. Ritter. 1987. Tributary entrenchment in response to lowered base level below a dam . Pp. 391-392 in: Erosion and Sedimentation in the Pacific Rim. International As sociation of Hydrological Sciences Publication 165. Gilbert, G.K. 1917. Hydraulic-mining debris in the Sierra Nevada. U.S. Geological Survey Profes sional Paper 105. 154 pp. Gillespie, P.O. 1981. Development of gully erosion at the head of Bango Creek near Yass, N.S. W. Journal of the Soil Conservation Service of New South Wales 37(1):6-12. Gleason, C.H. 1953. Indicators of erosion on watershed land in California. American Geophysical Union Transactions 34(3):419-426. 154 References

Gomez, 'B. 1983. Temporal vanatlOns in bedload transport rates: the effect of progressive bed armoring. Earth Surface Processes and Landforms 8(1):41-54. Gomez, B., and M. Church. 1989. An assessment of bed load sediment transport formulae for gravel bed rivers. Water Resources Research 25(6): 1161-1186. Gomez, B.; R.L Naff, and D.W. Hubbell. 1989. Temporal variations in bedload transport rates associated with the migration of bedforms. Earth Surface Processes and Landforms 14:135-156. Graf, W.H. 1971 . Hydraulics of Sediment Transport. McGraw-Hili, New York. 513 pp. Graf, W.H., and E.R. Acaroglu. 1968. Sediment transport in conveyance sy~tems (part I). Bulletin of the International Association of Scientific Hydrology 13(2). Grant, G. 1988. The RAPID technique: a new method for evaluating downstream effects of forest practices on riparian zones. U.S.D.A. Forest Service General Technical Report PNW-220. 36 pp. Griffen, D.M., Jf. 1979. Reservoir trap efficiency: the state of the art. Journal of Civil Engineering Design 1(4):355-377. Guy, H.P. 1969. Laboratory theory and methods for sediment analysis. Techniques of Water Resources Investigations of the U.S. Geological Survey. Book 5, Chapter CI. Guy, H.P., and V.W. Norman. 1970. Field methods for measurement of fluvial sediment. Tech niques of Water-Resources Investigations of the U.S. Geological Survey. Book 3, Chapter C2. Haible, W.W. 1980. Holocene profile changes along a California coastal stream. Earth Surface Processes 5(3):249-264. Hales, J.I.; A. Shindala, and K.H. Denson. 1970. Riverbed degradation prediction. Water Resources Research 6:549-556. Harden, D.R.; R.J. Janda, and K.M. Nolan. 1978. Mass movement and storms in the drainage basin of Redwood Creek, Humboldt County, California - a progress report. U.S. Geological Survey Open-file Report 78-486. 161 pp. Harr, R.D. 1981. Some characteristics and consequences of snowmelt during rainfall in western Oregon. Journal of Hydrology 53:277-304. Hassan, M.A. 1990. Scour, fill, and burial depth of coarse material in gravel bed streams. Earth Surface Processes and Landforms 15(4):341-356. Hassan, M.A.; M. Church, and A.P.S. Schick. 1991. Distance of movement of coarse particles in gravel bed streams. Water Resources Research 27:503-511. Hassan, M.A .; M. Church, and PJ. Ashworth. 1992. Virtual rate and mean distance of travel of individual clasts in gravel-bed channels. Earth Surface Processes and Landforms 17(6):617-627. Haupt, H.F. 1959. Road and slope characteristics affecting sediment movement from logging roads. Journal of Forestry 57:329-332. Heinemann, H.G. 1981. A new sediment trap efficiency curve for small reservoirs. Water Resources Bulletin 17(5). Herschy, R.W. 1985. Streamflow Measurement. Elsevier Applied Science Publishers, London. 553 pp. Hey, R.D. 1979. Flow resistance in gravel-bed rivers. Journal of the Hydraulics Division, American Society of Civil Engineers 105(HY4):365-379. Hollingshead, A.B. 1971. Sediment transport measurements in gravel river. Journal of the Hydrau lics Division, American Society of Civil Engineers 97(HY11): 1817-1834. Hooke, J.M. 1979. An analysis of the processes of bank erosion. Journal of Hydrology 42:39-62. Hooke, J.M. 1980. Magnitude and distribution of rates ofrier bank erosion. Earth Surface Processes 5(2): 143-157. References 155

Hubbell, D.W., and D.Q. Matejka. 1959. Investigations of sediment transportation, Middle Loup River at Dunning, Nebraska. u.s. Geological Survey Water Supply Paper 1476. Hulsing, H. 1967. Measurement of peak discharge at dams by indirect method. Chapter A5, Book 3, of Techniques of Water Resources Investigations of the U.S. Geological Survey. 29 pp. Hupp, C.R., and D.E. Bazemore. 1991. Dendrogeomorphic analysis of wetland sedimentation. Pp. 4- 40 to 4-47 in: Proceedings of the Fourth Federal Interagency Sedimentation Conference. 24-27 March 1986, Las Vegas, Nevada. Subcommittee on Sedimentation, Interagency Advisory Committee on Water Data. Imeson, A.C. 1976. Some effects of burrowing animals on slope processes in the Luxembourg Ardennes. Part 1: the excavation of animal mounds in experimental plots. Geografiska Annaler 58A(I-2): 115-125. Inglis, C.C. 1968. Discussion of: Systematic evaluation of river regime, by C.R. Neill and VJ. Galey. Journal of the Waterways and Harbors Division, American Society of Civil Engineers 94(WWI):109-114. Jackson, W.L. , and B.P. Van Haveren. 1987. Predicting channel responses to changing flow regimes: Beaver Creek, Alaska. Pp. 393-394 in : Erosion and Sedimentation in the Pacific Rim. International Association of Hydrological Sciences Publication 165. Jones, R.G.B., and M.A. Keech. 1966. Identifying and assessing problem areas in soil erosion surveys using aerial photographs. Photogram. Rec. 5: 189-197. Julien, P.Y., and M. Frenette. 1986. Scale effects in predicting soil erosion from large watersheds. Pp. 253-262 in R.F. Hadley (ed.): Drainage Basin Sediment Delivery. International Association of Hydrological Sciences Publication 159. Kalinske, A.A. 1947. Movement of sediment as bedload in rivers. Transactions, American Geo physical Union 28(4):615-620. Kalisz, PJ. 1986. Soil properties of steep Appalachian old fields. Ecology 67(4): 1011-1023. Keller, E.A., and F.J . Swanson. 1979. Effects of large organic material on channel form and fluvial processes. Earth Surface Processes 4:361-380. Kellerhals, R. , and D.J . Bray. 1971. Sampling procedures for coarse fluvial sediments. Journal of the Hydraulics Division, American Society of Civil Engineers 97: 1165-1180. Kelsey, H.M . 1980. A sediment budget and an analysis of geomorphic processes in the Van Duzen River basin, north coastal California, 1941-1975. Geological Society of America Bulletin Part 11,91:1119-1216. Kelsey, H.M. 1982. HilIslope evolution and sediment movement in a forested headwater basin, Van Duzen River, north coastal California. Pp. 86-96 in: Sediment Budgets and Routing in Forested Drainage Basins. U.S.D.A. Forest Service General Technical Report PNW-141. Kelsey, H.M.; R. Lamberson, and M.A. \I1adej. 1987. Stochastic model for the long-term transport of stored sediment in a river channel. Water Resources Research 23(9): 1738-1750. Khanbilvardi, R.M., and A.S. Rogowski. 1984. Quantitative evaluation of sediment delivery ratios. Water Resources Bulletin 20(6):865-874. Kirkby, MJ. 1987. The Hurst effect and its implications for extrapolating process rates. Earth Surface Processes and Landforms 12(1):57-68. Komar, P.D. 1980. Models of sediment transport in channelized water flows with ramifications to the erosion of the Martian outflow channels. Icarus 42:317-329. Komar, P.D. 1987a. Selective gravel entrainment and the empirical evaluation of flow competence. Sedimentology 34: 1165-1176. 156 References

Komar, P.O. 1987b. Selective grain entrainment by a current from a bed of mixed sizes: a reanaly sis. Journal of Sedimentary Petrology 57(2):203-211. Komar, P.O., and P.A. Carling. 1991. Grain sorting in gravel-bed streams and the choice of particle sizes for flow-competence evaluations. Sedimentology 38(3):489-502. Komar, P.O., and Z. Li. 1986. Pivoting analysis of the selective entrainment of sediments by shape and size with application to gravel threshold. Sedimentology 33 :425-436. Kotarba, A. 1986. Soil creep in semi-arid zone of Central Mongolia. Zeitschrift flir Geomorphologie Neue Folge Supplementband 60:245-249. Kreznor, W.R.; K.R. Olson, D.L. Johnson, and R.L. Jones. 1990. Quantification of postsettlement deposition in a northwestern Illinois sediment basin. Soil Science Society of America Journal 54(5): 1393-140 I. Kuhnle, R.A. ; J.e. Willis, and A.J. Bowie. 1989. Total sediment load calculations for Goodwin Creek. pp. 700-705 in: S.Y. Wang (ed.), Sediment Transport Modeling. American Society of Civil Engineers, New York, N.Y. 829 pp. Laird, J.R., and M.D. Harvey. 1986. Complex-response of a chaparral drainage basin to fire. Pp. 165-183 in R.F . Hadley (ed .): Drainage Basin Sediment Delivery. International Association of Hydrological Sciences Publication 159. 487 pp. LaMarche, V.C. 1966. An 800-year history of stream erosion as indicated by botanical evidence. u.S. Geological Survey Professional Paper 550-0. Pp. 83-86. Lane, L.J.; D.L. Schertz, E.E. Alberts, J.M. Laflen, and V.L. Lopes. 1988. The U.S. national project to develop improved erosion prediction technology to replace the USLE. Pp. 473-482 in : Inter national Association of Hydrological Sciences Publication 174. Lara, 1.M., and E.L. Pemberton. 1963. Initial unit weight of deposited .sediments. Pp. 818-845 in: Proceedings of the Federal Interagency Sedimentation Conference. U.S.D.A . Miscellaneous Publication 970. Laronne, 1.B., and M.A. Carson. 1976. Interrelationships between bed morphology and bed-material transport for a small, gravel-bed channel. Sedimentology 23:67-85. Larson, K.R., and R.C. Sidle. 1980. Erosion and sedimentation data catalog of the Pacific North west. U .S.D.A. Forest Service, Pacific Northwest Region ; R6-WM-050-1981; 64 pp. Laursen, E.M. 1958. The total sediment load of streams. Journal of the Hydraulics Division, American Society of Civil Engineers 54(HYI):1-36. Lehre, A.K. 1982. Sediment budget of a small Coast Range drainage basin in North-Central California. Pp. 67-77 in : Sediment Budgets and Routing in Forested Drainage Basins. U.S.D.A. Forest Service General Technical Report PNW -141. Lehre, A.K., B.D. Collins, and T. Dunne. 1983 . Post-eruption sediment budget for the North Fork Toutle River Drainage, June 1980-June 1981. Zeitschrift flir Geomorphologie Neue Folge Sup plementband 46:143-163. Leopold, L.B.; W.W. Emmett, and R.M. Myrick. 1966. Channel and hillslope processes in a semiarid area, New Mexico. U.S. Geological Survey Professional Paper 352G: 193-253. Leopold, L.B. ; M.G. Wolman, and J.P. Miller. 1964. Fluvial Processes in Geomorphology. W.H. Freeman Co., San Francisco. 522 pp. Lewis, D.T., and M.J. Lepele. 1982. Quantification of soil loss and sediment produced from eroded land. Soil Science Society of America Journal 46:369-.372. Likens, G.E., and R.E. Bilby. 1982. Development, maintenance, and role of organic-debris dams in New England streams. Pp. 122-128 in: Sediment Budgets and Sediment Routing in Forested Drainage Basins. U.S. Forest Service General Technical Report PNW-141. References 157

Love, D. W. 1986. A geological perspective of sediment storage and delivery along the Rio Pl,lerco, central New Mexico. Pp. 305-322 in R.F. Hadley (ed.): Drainage Basin Sediment Delivery. In ternational Association of Hydrological Sciences Publication 159.487 pp. Lyles, L. 1983. Erosive wind energy distributions and climatic factors for the West. Journal of Soil and Water Conservation 38(2): I 06-1 09. Lyles, L., and B.E. Allison. 1980. Range grasses and their small grain equivalents for wind erosion control. Journal of Range Management 33(2):143-146. Maddock, T., Jr. 1976. Equations for resistance to flow and sediment transport in alluvial channels. Water Resources Research 12(1): 11-21. Maddock, T., Jr. 1983. Discussion: Sediment transport and unit stream power function. Journal of the Hydraulics Division, American Society of Civil Engineers I 09( 12): 1779-1781. Madej, M.A. 1982. Sediment transport and channel changes in an aggrading stream in the Puget Lowland, Washington. Pp. 97-108 in: Sediment Budgets and Sediment Routing in Forested Drainage Basins. U.S.D.A. Forest Service General Technical Report PNW-141. Madej, M.A. 1987. Residence times of channel-stored sediment in Redwood Creek, northwestern California. Pp. 429-438 in Erosion and Sedimentation in the Pacific Rim. International Asso ciation of Hydrological Sciences Publication 165. Maeglin, R.R. 1979. Increment cores - how to collect, handle, and use them. U.S.D.A. Forest Service General Technical Report FPL-25. 18 pp. Mahmood, K. 1980. Verification of sediment transport functions-Missouri River. U.S. Army Engineer Division, Missouri River. Corps of Engineers, Omaha, Nebraska. MRD Sediment Series no. 19. Maner, S.B. 1953 . Factors affecting sediment delivery ratio in the Red Hills physiographic area. Transactions, American Geophysical Union 39(4):669-675. Mark, D.M. 1974. Line intersection method for estimating drainage density. Geology 2(5):235-236. Maser, C.; R.F. Tarrant, J.M. Trappe, and 1.F. Franklin (eds.). 1988. From the forest to the sea: a story offallen trees. U.S.D.A. Forest Service General Technical Report PNW-229. Matherne, A.M., and K.L. Prestegaard. 1988. Hydrological characteristics as a determinant of sediment delivery in watersheds. Pp. 89-96 in: Sediment Budgets. International Association of Hydrological Sciences Publication 174. Matthai, H.F. 1967. Measurement of peak discharge at width contractions by indirect methods. Chapter A4, Book 3, of Techniques of Water Resources Investigations of the U.S. Geological Survey. 44 pp. McGuinness, J.L.; L.L. Harrold, and W.M. Edwards. 1971. Relation of rainfall energy and stream flow to sediment yield from small and large watersheds. Journal of Soil and Water Conservation 26:233-235. McHarg,I.L. 1969. Design with Nature. Natural History Press, New York. 198 pp. McLean, D.G., and B.L. Tassone. 1991. A sediment budget of the lower Fraser River. Pp. 2-33 to 2- 40 in: Proceedings of the Fifth Federal Interagency Sedimentation Conference, March 18-21 , 1991, Las Vegas, Nevada. Federal Energy Regulatory Comm ission. McNeil, W.J., and W.H. Ahnell. 1960. Measurement of gravel composition of salmon stream beds. Fisheries Research Institute, University of Washington, Seattle, WA. Circular 120.7 pp. Megahan, W.F. 1974. Erosion over time on severely disturbed granitic soils: a model. U.S.D.A. Forest Service Research Paper INT-156. 14 pp. Megahan, W.F. 1975 . Sedimentation in relation to logging activities in the mountains of central Idaho. In: Present Prospective Technology for Predicting Sediment Yields and Sources, 158 References

Proceedings of the Sediment Yield Workshop, U.S.D.A. Sedimentation Laboratory, Oxford, Mississippi. Nov 28-30, 1972. U.S Agricultural Research Service, ARS-S-40. Pp. 74-82. Megahan, W.F. ; K.A. Seyedbagheri, and P.C. Dodson. 1983. Long-term erosion on granitic roadcuts based on exposed tree roots. Earth Surface Processes and Landforms 8(1): 19-28. Megahan, W.F.; K.A. Seyedbagheri, T.L. Mosko, and G.L. Ketcheson. 1986. Construction phase sediment budget for forest roads on granitic slopes in Idaho. Pp. 31-39 in: Drainage Basin Sediment Delivery. International Association of Hydrological Sciences Publication 159. Milhous, R.T. 1973. Sediment transport in a gravel-bottomed stream. Ph.D. Bissertation, Oregon State University, Corvallis, Oregon. Miller, J.R., and S.G. Wells. 1986. Types and processes of short-term sediment and uranium-tailings storage in arroyos: an example from the Rio Puerco of the West, New Mexico. Pp. 335-353 in R.F. Hadley (ed.): Drainage Basin Sediment Delivery. International Association of Hydrologi cal Sciences Publication 159.487 pp. Miller, M.C .; LN. McCave, and P.O. Komar. 1977. Threshold of sediment motion in unidirectional currents. Sedimentology 24:507-528. Mizumara, K. 1989. Hydrologic approach to prediction of sediment yield. Journal of Hydraulic Engineering 115(4):529-535. Moglen, G.E., and R.H. McCuen. 1988. Effects of detention basins on in-stream sediment move ment. Journal of Hydrology 104(1-4):129-140. Moore, 1.0., and G.J. Burch. 1986. Sediment transport capacity of sheet and rill flow: application of unit stream power theory. Water Resources Research 22(8): 1350-1360. Mosley, M.P. 1981. The influence of organic debris on channel morphology and bedload transport in Ii New Zealand forest stream. Earth Surface Processes and Landforms 6(6):571-580. Mosley, M.P., and D.S. Tindale. 1985. Sediment variability and bed material sampling in gravel-bed rivers. Earth Surface Processes and Landforms 10(5):465-482. Nakato, T. 1990. Tests of selected sediment-transport formulas. Journal of Hydraulic Engineering 116:362-379. NCASL 1985. Catalog of landslide inventories for the Northwest. National Council of the Paper Industry for Air and Stream Improvement Technical Bulletin 456. 78 pp. Nearing, M.A.; G.R. Foster, LJ. Lane, and S.c. Finkner. 1989. A process-based soil erosion model for U.S.D.A. - Water Erosion Prediction Project Technology. Transactions of the American Society of Agricultural Engineers 32(5): 1587-1593. Nelson, L.M. 1971. Sediment transport by streams in the Snohomish River Basin, Washington, October 1967-July 1969. U.S. Geological Survey Open File Report. Tacoma, WA. 44 pp. Nilsen, T.H.; F.A. Taylor, and E.E. Brabb. 1976. Recentlandslides in Alameda County, California (1940-71): an estimate of economic losses and correlations with slope, rainfall, and ancient landslide deposits. U.S. Geological Survey Bulletin 1398,21 pp. Norton, L.D. 1986. Erosion-sedimentation in a closed drainage basin in northwest Indiana. Soil Science Society of America Journal 50:209-231. Norton-Griffiths, M. 1983 . Application of low level aerial survey to studies of soil erosion. Pp. 123- 150 in: Soil and Water Conservation in Kenya, Proceedings of a workshop. University of Nairobi, Kenya. O'Loughlin, C.L. 1972. An investigation of the stability of the steepland forest soils in the Coast Mountains, Southwest British Columbia. Ph.D. Dissertation, University of British Columbia, 147 pp. References 159

Osterkamp, W.R., and R.S. Parker. 1991. Sediment monitoring in the United States. Pp. 1-15 to 1-24 in : Proceedings of the Fifth Federal Interagency Sedimentation Conference, March 18-21, 1991, Las Vegas, Nevada. Federal Energy Regulatory Commission. Parker, G. 1990. Surface-based bedload transport relation for gravel rivers. Journal of Hydraulic Research 28(4):417-436. Parker, G., and P.C. Klingeman. 1982. On why gravel bed streams are paved. Water Resources Research 18:1409-1423. Parker, G.; P.C . Klingeman, and D.C. McLean. 1982. Bedload and size distribution in paved gravel bed streams. Journal of the Hydraulics Division, American Society of Civil Engineers 108:545- 571. Parker, R.S. 1988. Uncertainties in defining the suspended sediment budget for large drainage basins. Pp. 523-532 in: International Association of Hydrological Sciences Publication 174. Pearce, AJ., and A. Watson. 1983. Medium-term effects of two landsliding episodes on channel storage of sediment. Earth Surface Processes and Landforms 8(1 ):29-40. Pemberton, E.L. 1975. Channel changes in the Colorado River below Glen Canyon Dam. Pp. 5-61 to 5-73 in: Proceedings of the Third Federal Inter-Agency Sedimentation Conference, March 22-25, 1976, Denver CO. Sedimentation Committee, Water Resources Council. Pp. 5-25 to 5- 36. Perkins, S.J. 1989. Landslide deposits in low-order streams - their erosion rates and effects on channel morphology. Pp. 173-182 in: Headwaters Hydrology. American Water Resources Association. Phillips, J.D. 1990. Fluvial sediment budgets in the North Carolina Piedmont. Geomorphology 4(3/4):231-242. Phipps, R.L. 1974. The soil creep - curved tree fallacy. Journal of Research of the U.S. Geological Survey 2(3):371-377. . Pickup, G., and RJ. Higgins. 1979. Estimating sediment transport in a braided gravel channel - the Kawerong River, Papua New Guinea. Journal of Hydrology 40:283-297. Pizzuto, J.E., and T.S. Meckelnburg. 1989. Evaluation of a linear bank erosion equation. Water Resources Research 25(5): I 005-1 0 13. Pole, M.W., and D.R. Satterlund. 1978. Plant indicators of slope instability. Journal of Soil and Water Conservation 33(5):230-232. Potter, K.N.; T.M. Zobeck, and LJ. Hagan. 1990. A microrelief index to estimate soil erodibility by wind. Transactions of the American Society of Agricultural Engineers 33( I): 151-155. Prestegaard, K.L. 1988. Morphological controls on sediment delivery pathways. Pp. 533-540 in: International Association of Hydrological Sciences Publication 174. Rango, A. 1970. Possible effects of precipitation modification on stream channel geometry and sediment yield. Water Resources Research 6: 1765-1770. Rapp, A. 1960. Recent developments of mountain slopes in Karkevagge and surroundings, northern Scandinavia. Geografiska Annaler 42:71-200. Reid, I.; L.E. Frostick, and J.T. Layman. 1985. The incidence and nature of bedload transport during flood flows in coarse-grained alluvial channels. Earth Surface Processes and Landforms 10:33- 44. Reid, L.M. 1981. Sediment production from gravel-surfaced forest roads, Clearwater basin, Washington. University of Washington Fisheries Research Institute Publication FRI-UW-Sl OS. 247 pp. 160 References

Reid, L.M. 1982. Rates and causes of soil erosion in southeast Shinyanga Region, Tanzania. Southeast Shinyanga Land Use Study, Volume 3. EcoSystems, Ltd., Nairobi, Kenya. Report to the Shinyanga Rural Integrated Development Programme. 200 pp. Reid, L.M. 1983. A reconnaissance survey of erosion processes in Machakos District, Kenya. Pp. 105-122 in: Soil and Water Conservation in Kenya. Proceedings of a workshop, University of Nairobi, Kenya. Reid, L.M. 1989. Channel incision by surface runoff in grassland catchments. Ph.D. Dissertation, University of Washington, Seattle. Reid, L.M. 1990. Construction of sediment budgets to assess erosion in Shinyanga Region, Tanza nia. Pp. 105-114 in: Research Needs and Applications to Reduce Erosion and Sedimentation in Tropical Steeplands. International AssQciation of Hydrological Sciences Publication 192. 396 pp. Reid, L.M. 1993. Research and cumulative watershed effects. U.S.D.A. Forest Service General Technical Report, Pacific Southwest Research Station, Albany, California, GTR-PSW-141. 118 p. Reid, L.M., and T. Dunne. 1984. Sediment production from forest road surfaces. Water Resources Research 20(11): 1753-1761. . Reid, L.M.; T. Dunne, and C.J. Cederholm. 1981. Application of sediment budget studies to the evaluation of logging road impact. Journal of Hydrology (New Zealand) 20:49-62. Reid, L.M.; T. Dunne, and J. Lewis. In review. Calculating average landslide frequency: a simplified method. Reid, L.M., and C.W. Smith. 1992. The effects of Hurricane Iniki on flood hazard on Kauai. USDA Forest Service Pacific Southwest Research Station. Report to the Hawaii Department of Land and Natural Resources, the USDA Forest Service, and the US Soil Conservation Service. 26 p. Reneau, S.L., and W.E. Dietrich. 1991. Erosion rates in the southern Oregon Coast Range: evidence for an equilibrium between hillslope erosion and sediment yield. Earth Surface Processes and Landforms 16(4):307-322. Rice, R.M., and N.H. Pillsbury. 1982. Predicting landslides in clearcut patches. Pp. 303-311 in: Recent Developments in the Explanation and Prediction of Erosion and Sediment Yield. Inter national Association of Hydrological Sciences Publication 137. Richards, K. 1990. Fluvial geomorphology: initial motion of bed material in gravel-bed rivers. Progress in Physical Geography 14(3):395-415. Riggs, H.C. 1976. A simplified slope-area method for estimating flood discharges in channels. Journal of Research of the U.S. Geological Survey 4:285-291. Rijn, L.C. van. 1984a. Sediment transport, part I: Bed load transport. Journal of Hydraulic Engi neering 110(10): 1431-1456. Rijn, L.C. van. 1984b. Sediment transport, part II: Suspended load transport. Journal of Hydraulic Engineering 11O( 11): 1613-1641. Roberts, R.G., and M. Church. 1986. The sediment budget in severely disturbed watersheds, Queen Charlotte Ranges, British Columbia. Canadian Journal of Forest Research 16: 1092-11 06. Roehl, J. W. 1962. Sediment source areas, delivery ratios, and influencing morphological factors. Pp. 202-213 in: International Association of Scientific Hydrology Commission on Land Erosion, PuMication 59. Roels, J.M. 1985. Estimation of soil loss at a regional scale based on plot measurements - some critical considerations. Earth Surface Processes and Landforms 10(6):587-598. Rottner, J. 1959. A formula for bed load transportation. La Houille Blanche 4(3):301-307. References 161

Ryan, S.E., and G.E. Grant. 1991. Downstream effects of timber harvesting on channel morphology in Elk River Basin, Oregon. Journal of Environmental Quality 20( I ):60-72. Santantonio, D.; R.K. Hermann, and W.S. Overton. 1977. Root biomass studies in forest ecosys tems. Pedobiologia 17( I): 1-31. Saunders, I., and A. Young. 1983. Rates of surface processes on slopes, slope retreat, and denuda tion. Earth Surface Processes and Landforms 8:473-501. Savabi, M.R., and G.F. Gifford. 1987. Application of selected soil loss equations to trampled soil conditions. Water Resources Bulletin 23(4):709ff. Schaetzl, R.J .; OJ. Johnson, S.F. Burns, and T.W. Small. 1989. Tree uprooting: review ofterminol ogy, process, and environmental implications. Canadian Journal of Forest Research 19(1): I-II. Seginer, I. 1966. Gully development and sediment yield. Journal of Hydrology 4(3):236-252. Shen, H.W., and C.S. Hung. 1972. An engineering approach to total bed-material load by regression analysis. Pp. 14-1 to 14-17 in H.W. Shen (ed.): Sedimentation. Published by H.W. Shen, Fort Collins, Colorado. Shen, H.W., and C.S. Hung. 1983. Remodified Einstein procedure for sediment load. Journal of Hydraulic Engineering 109(4):565-578. Shields, A. 1936. Applicatioon of similarity principles and turbulence research to bedload measure ment. California Institute of Technology, Pasadena, CA. Report 167. Sidle, R.C. 1988. Bedload transport regime of a small forest stream. Water Resources Research 24(2):207-218. Sigafoos, R.S. 1964. Botanical evidence of floods and flood-plain deposition. U.S. Geological Survey Professional Paper 485A. Simonett, D.S. 1967. Landslide distribution and earthquakes in the Bewani and Torricelli mountains, New Guinea. Pp. 64-84 in 1.N. Jennings and 1.A. Mabbutt (eds.): Landform Studies from Australia and New Guinea. Australian National University Press, Canberra. Sing, E.F. 1986. River sediment budget. Proceedings of the 4th Interagency Sedimentation Confer ence, Pp. 3-73 to 3-82 in: Proceedings of the Fourth Federal Interagency Sedimentation Confer ence. 24-27 March 1986, Las Vegas, Nevada. Subcommittee on Sedimentation, Interagency Advisory Committee on Water Data. Skidmore, E.L., and N.P. Woodruff. 1968. Wind erosion forces in the United States and their use in predicting soil loss. U.S.D.A. Agricultural Research Service, Agricultural Handbook 346. 42 pp. Slosson, J.E. 1979. Recognition of landslides. Pp. 113-127 in: D. Buesch (ed.), Geomorphic Applications in Engineering Geology. Short course held at Cal. State University Los Angeles, Nov. 10-11, 1979. Reprinted by the Southern California Section of the Association of Engi neering Geologists. Smith, N.D. 1974. Sedimentology and bar formation in the Upper Kicking Horse River, a braided outwash stream. Journal of Geology 82:205-223. Smith, R.D., and F.J. Swanson. 1987. Sediment routing in a small drainage basin in the blast zone at Mount St. Helens, Washington, USA. Geomorphology 1: 1- 13. Spaulding, P., and J.R. Hansbrough. 1944. Decay of logging slash in the Northeast. U.S.D.A. Technical Bulletin 876. Stall, 1.B.; N.L. Rupani, and P.K. Kandaswamy. 1958. Sediment transport in Money Creek. Proceedings of the American Society of Civil Engineers 84(HYI). Stevens, H.H., Jr., and c.T. Yang. 1989. Summary and use of selected fluvial sediment-discharge formulas. U.S. Geological Survey Water Resources Investigations Report 89-4026. 162 References

Stott, T.A.; R.1. Ferguson, R.C. Johnson, and M.D. Newson. 1986. Sediment budgets in forested and un forested basins in upland Scotland. Pp. 57-68 in R.F. Hadley (ed.): Drainage Basin Sediment Delivery. International Association of Hydrological Sciences Publication 159.487 pp. Sutherland, R.A., and R.B. Bryan. 1991. Sediment budgeting: a case study in the Katiorin drainage basin, Kenya. Earth Surface Processes and Landfonns 16(5):383-398. Swanson, F.1., and C.T. Dyrness. 1975. Impact of clearcutting and road construction on soil erosion by landslides in the western Cascade Range, Oregon. Geology 3:393-396. Swanson, F.1 ., and R.L. Fredriksen. 1982. Sediment routing and budgets: implications for judging impacts of forestry practices. Pp. 129-137 in: Sediment Budgets and Routing in Forested Drain age Basins. u.s.D.A. Forest Service General Technical Report PNW-141. Swanson, F.J., R.L. Fredriksen, and F.M. McCorison. 1982. Material transfer in a western Oregon forested watershed. Pp. 233-266 in: Analysis of Coniferous Forest Ecosystems in the western United States (R.L. Edmonds, ed.). Hutchinson, Dowden, Ross Pub. Co., Stroudsburg, PA. Swift, L.W., Jr. 1984. Soil losses from roadbeds and cut and fill slopes in the southern Appalachian Mountains. Southern Journal of Applied Forestry 8(4):209-215. Thompson, J.R. 1964. Quantitative effect of watershed variables on rate of gully head advancement. American Society of Agricultural Engineers Transactions 7:54-55. Thompson, S.M. 1985. Transport of gravel by flows up to 500 m3s- l, Ohau River, Otago, New Zealand. Journal of Hydraulic Research 23:285-303. Thorn, C.E. 1978. A preliminary assessment of the geomorphic role of pocket gophers in the alpine zone of the Colorado Front Range. Geografiska Annaler 60A(3-4): 181-187. Thome, C.R., and L. W. Zevenbergen. 1985. Estimating mean velocity in mountain rivers. Journal of Hydraulic Engineering 111(4):612-624. Toffaleti, F.B. 1969. Definitive computations of sand discharge in rivers. Journal of the Hydraulics Division, American Society of Civil Engineers 95(HY1):225-248. Tollner, E.W.; B.1. Barfield, C.T. Haan, and T.Y. Kao. 1976. Suspended sediment filtration capacity of rigid vegetation. Transactions of the American Society of Agricultural Engineers 19(4):678- 682. Trimble, G.R., and R.S. Sartz. 1957. How far from a stream should a logging road be located? Journal of Forestry 55:339-341. Trimble, S.W. 1981. Changes in sediment storage in the Coon Creek basin, Driftless Area, Wiscon sin, 1853 to 1975. Science 214:181-183. Trimble, S.W. 1983. A sediment budget for Coon Creek basin in the Driftless Area, Wiscinsin, 1853 to 1977. American Journal of Science 283:454-474. UNEP. 1978. Report on the FAOIUNEP expert consultation on methodology for assessing soil degradation. Rome, 25-27 January, 1978. United Nations Food and Agriculture Organization, United Nations Environmental Programme. USGS. 1989. USGS Water Resources Data - California, Water Year 1988, Vol. 4. U.S. Geological Survey Water-Data Report CA-88-4. US Soil Conservation Service. 1977. Procedure for detennining rates of land damage, land depreciation and volume of sediment produced by gully erosion. Pp. 125-141 in : S.H. Kunkle and J.L. Thames (editors), Guidelines for Watershed Management. UN Food and Agriculture Organization, Rome. FAO Conservation Guide 1. Van der Plas, L., and A.C. Tobi. 1965. A chart for judging the reliabiHty of point counting results. American Journal of Science 263:87-90. References 163

Vanoni, V.A. (ed.). 1975. Sedimentation Engineering. American Society of Civil Engineers, New York, NY. 745 pp. Vanoni, V.A.; N.H. Brooks, and J.F. Kennedy. 1961. Lecture notes on sediment transportation and channel stability. W.M. Keck Laboratory, California Institute of Technology, Pasadena, CA. Report no. KH-R I, September 1960, with minor corrections and additions, January, 1961. Van Sickle, J. 1981. Long-term distributions of annual sediment yields from small watersheds. Water Resources Research 17(3):659-663. Van Sickle, J., and R.L. Beschta. 1983. Supply-based models of suspended sediment transport in streams. Water Resources Research 19(3):768-778. Voslamber, B., and A.W.L. Veen. 1985. Digging by badgers and rabbits on some wooded slopes in Belgium. Earth Surface Processes and Landforms 10(1 ):79-82. Walkotten, W.J. 1976. An improved technique for sampling streambed sediments. U.S.D.A. Forest Service Research Note PNW-28I. 11 pp. Walling, D.E.; S.B. Bradley, and c.P. Lambert. 1986. Conveyance losses of suspended sediment within a flood plain system. Pp. 119-131 in R.F. Hadley (ed.): Drainage Basin Sediment Deliv ery. International Association of Hydrological Sciences Publication 159.487 pp. Walter, T. 1985. Prairie gully erosion in the Redwood Creek basin, California. Redwood National Park Research and Development Technical Report 16.24 pp. Warburton, J. 1990. Comparison of bed load yield estimates for a glacial meltwater stream. Pp. 315- 323 in: Hydrology in Mountain Regions I-Hydrological Measurements; the Water Cycle. International Association of Hydrological Sciences Publication 193. Way, D.S. 1978. Terrain Analysis. Dowden, Hutchinson and Ross. Stroudsburg PA. 438 pp. White, R.S. 1986. Sediment routing - an electronic spreadsheet template. P. 6-34 to 6-43 in: Proceedings of the Fourth Federal Interagency Sedimentation Conference. 24-27 March 1986, Las Vegas, Nevada. Subcommittee on Sedimentation, Interagency Advisory Committee on Water Data. White, W.R.; H. Milli, and A.D. Crabbe. 1978. Sediment transport; an appraisal of available methods. UK Hydraulics Research Station Report 119. Wallingford, England. Wiberg, P.L., and J.D. Smith. 1987. Calculations of the critical shear stress for motion of uniform and heterogeneous sediments. Water Resources Research 23: 1471-1480. Wieczoreck, G.F. 1984. Preparing a detailed landslide-inventory map for hazard evaluation and reduction. Bulletin of the Association of Engineering Geologists 21 (3):337-342. Williams, G.P. 1978. Bankfull discharge of rivers. Water Resources Research 14(6): 1141-1154. Williams, G.P. 1983. Improper use ofregression equations in earth sciences. Geology II: 195-197. Williams, lR. 1975. Sediment routing for agricultural watersheds. Water Resources Bulletin 11(5):965-974. Williams, J.R., and H.D. Berndt. 1972. Sediment yield computed with universal equation. Journal of the Hydraulics Division, American Society of Civil Engineers 98(HYI2):2087-2097. Williams, T.H.L. 1981. An error analysis of the photographic technique for measuring percent vegetative cover. Soil Science Society of America Journal 43(3):578-582. Wilson, D. 1985. Subjective techniques for identification and hazard assessment of unstable terrain. Pp. 36-42 in: Proceedings of the Workshop on slope stability: problems and solutions in forest management (D. Swanston, ed.). U.S.D.A. Forest Service General Technical Report PNW-180. Winkworth, R.E.; R.A. Perry, and C.O. Rossetti. 1962. A comparison of methods of estimating plant cover in an arid grassland community. Journal of Range Management 15(4): 194-196. 164 References

Wischmeier, W.H. 1976. Use and misuse of the Universal Soil Loss Equation. Journal of Soil and Water Conservation 31 (I ):5-9. Wischmeier, W.H., and L.D. Meyer. 1973 . Soil erodibility on construction sites; in Soil Erosion: causes, mechanisms, prediction, and control. Highway Research Board Special Report 135, pp. 20-29. Wischmeier, W.H., and D.D. Smith. 1978. Predicting rainfall erosion losses-a guide to conservation planning. U.S.D.A. Agricultural Handbook 537. 58 pp. Wolman, M.G. 1954. A method of sampling coarse bed material. Transactions, American Geophysi cal Union 35:95 I -956. Wolman, M.G . 1959. Factors influencing erosion of a cohesive river bank. American Journal of Science 257:204-216. Yalin, M.S. 1963. An expression for bed-load transportation. Journal of the Hydraulics Division, American Society of Civil Engineers 89(HY3):222-250. Yang, C.T. 1973. Incipient motion and sediment transport. Journal of the Hydraulics Division, American Society of Civil Engineers 99(HY 10) : 1679-1704. Yang, c.T. 1979. Unit stream power equations for total load. Journal of Hydrology 40(1 /2): 123-138. Yang, C.T. 1984. Unit stream power equation for gravel. Journal of Hydraulic Engineering 110:1783-1797. Yang, C.T., and A. Molinas. 1982. Sediment transport and unit stream power function. Journal of the Hydraulics Division, American Society of Civil Engineers 108(HY6):774-793. Yang, C.T., and J.B. Stall. 1976. Applicability of unit stream power equation. Journal of the Hydraulics Division, American Society of Civil Engineers 102(5):559-568. Young, M.K.; W.A. Hubert, and T.A. Wesche. 1991. Biases associa~ed with four stream substrate samplers. Canadian Journal of Fisheries and Aquatic Sciences 48: 1882-1886. Zachar, D. 1982. Soil Erosion. Elsevier, Amsterdam.

Addresses of authors: Leslie M. Reid USDA Forst Service Pacific Southwest Research Station Arcata, California USA Thomas Dunne School of Environmental Sciences and Management University of California Santa Barbara, California USA

View publication stats